Base_JSON/prefixJ/json/J77/J77-1006.json

{
    "paper_id": "J77-1006",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T02:49:41.750224Z"
    },
    "title": "E X P L A l A T I O N C A P A B I L I T I E S OF PRODUCTION -BASED CONSULTATION SYSTEMS",
    "authors": [
        {
            "first": "A",
            "middle": [
                "Carlisle"
            ],
            "last": "Scott",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "University",
                "location": {
                    "addrLine": "J u l y 1976"
                }
            },
            "email": ""
        },
        {
            "first": "William",
            "middle": [
                "J"
            ],
            "last": "Clancey",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "University",
                "location": {
                    "addrLine": "J u l y 1976"
                }
            },
            "email": ""
        },
        {
            "first": "Randall",
            "middle": [],
            "last": "Davis",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "University",
                "location": {
                    "addrLine": "J u l y 1976"
                }
            },
            "email": ""
        },
        {
            "first": "Edward",
            "middle": [
                "H"
            ],
            "last": "Shortliffe",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "University",
                "location": {
                    "addrLine": "J u l y 1976"
                }
            },
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "Research sponsored in p a r t by Bureau of Health Services Research and Evaluation, Grant HS01544; Biotechnology eResources, Grant RR-00785; Advanced Research Projects Agency, ARPA Contract DAHC-15-73-C-0435; National Institute of Health, Grant GM-01922. Copyrzght Q 1977 Association for Computational G i n g u i s t i c s 131 Davis, R., Suchanan, B. G., and S h o r t l i f fe, E. H. Production Rules as an Approach t o Knowledqe-Based Consultation Systems. A 1 Memo 266, S t a n f o r d A r t i f i c i a l I n t e l l i g e n c e Laboratory, S t a n f o r d U n i v e r s i t y , October 1975. Also a c c e p t e d for p u b l i c a t i o n i n Artificial I n t e l l i g e n c e , Februarv 1977, [4] Davis, R. Applications of Meta Level Knowledge t o the C o n s t r u c t ion, Maintenance and Use of Large Knowledqe Bases. D o c t o r a l d i s s e r t a t i o n , Stanford University, June 1976. Also available as A 1 Memo 283, Stanford",
    "pdf_parse": {
        "paper_id": "J77-1006",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "Research sponsored in p a r t by Bureau of Health Services Research and Evaluation, Grant HS01544; Biotechnology eResources, Grant RR-00785; Advanced Research Projects Agency, ARPA Contract DAHC-15-73-C-0435; National Institute of Health, Grant GM-01922. Copyrzght Q 1977 Association for Computational G i n g u i s t i c s 131 Davis, R., Suchanan, B. G., and S h o r t l i f fe, E. H. Production Rules as an Approach t o Knowledqe-Based Consultation Systems. A 1 Memo 266, S t a n f o r d A r t i f i c i a l I n t e l l i g e n c e Laboratory, S t a n f o r d U n i v e r s i t y , October 1975. Also a c c e p t e d for p u b l i c a t i o n i n Artificial I n t e l l i g e n c e , Februarv 1977, [4] Davis, R. Applications of Meta Level Knowledge t o the C o n s t r u c t ion, Maintenance and Use of Large Knowledqe Bases. D o c t o r a l d i s s e r t a t i o n , Stanford University, June 1976. Also available as A 1 Memo 283, Stanford",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "A computer program t h a t models an expert in a g i v e n domain f a more l i k e l y to be accepted by experts i n t h a t domain, and by non-experts seeking i t 9 advice, i f the system c a n explain i t s actions.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "An ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "Scope of MYCIN'S E x p l a n a t i o n Capability . . rn . . program-generated e x~l a n a t i o n s . ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "2.3",
                "sec_num": null
            },
            {
                "text": "The purpose of an explanation c a~a b i l i t v (KC) i s t o qive the user access t o a s much of the system s knowledge as posbible. I d e a l l y , i t should be easy for a user t o get a complete, understandable answer t o anv sort of question about t h e system's knowledge and pera at ion -both i n general, and w i t h reference t o a particular consultation. This l l p l i e s three major goals i n t h e development of an explanation capability. ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Performance Characteristics of an Explanation Capability",
                "sec_num": null
            },
            {
                "text": "What decision ~t made about aome subproblem Why i t didn t use a certain plece of information Why i t failed t o ma!ce a certain decision Why it required a certain piece o f infornatlon Why it didn t require a certain ~i e c e of information How i t w i l l find out a certain plece of Information [while tRe consultatinn i s i n oroqress] What the system i s currently doing? [while t h e consultation is i n ~r o a r e s s ] The m e c i f i c s e t of explanation types which are chosen as basics, however, w i l l depend on the particular system.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "How it made a certain decision How i t used a piece of information",
                "sec_num": null
            },
            {
                "text": "2) To enable t h e user t o get an explanation which answers t h e question completely and comprehensively.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "How it made a certain decision How i t used a piece of information",
                "sec_num": null
            },
            {
                "text": "3) To make the EC easy t o use. A novice should be able t o use t h e EC without f i r s t spending a large amount of time lehrninq how t o request explanations. r e a s o n i n g -s t a t u s checker, a n d i t s g e n e r a l . q u e s t i o n -a n s w e r e r w i l l o n l y be a c c e s s i b l e a t t h e t e r m i n a t i o n of the c o n s u l t a t i a n .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "How it made a certain decision How i t used a piece of information",
                "sec_num": null
            },
            {
                "text": "An EC must know what is i n t h e system's knowledge base, and how it r u l e s Also may be necessary t o t h e extent t h a t t h e c o n t e n t of a r u l e may e x p l a i n why it was necessary to u s e t h i s rule, or may affect which futupe rules w i l l be t r l e d .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Knowledge Requirements of an Explanation Capability",
                "sec_num": null
            },
            {
                "text": "A g e n e r a l question-answerer will need more ~n f o m a t l o n about t h e system since t h e scope o f i t s e x p l a n a t i o n s is much broader. i t s t a s k is t o answer g e n e r a l questions about Lhe system's knowledge base. To do this, i t must know how the system stores knowledge about i t s area of e x p e r t i s e ( t h e static kdowledge with which it starts each c o n s u l t a t i o n ) and how i t stores facts gathered d u r i n g a p a r t i c u l a r c o n s u l t a t i o n (its dynamlc knowledge).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Knowledge Requirements of an Explanation Capability",
                "sec_num": null
            },
            {
                "text": "These two types o f information will a l l o w a GQA t o answer g u e s t i o n s about t h e substance and e x t e n t of the ~r o d u c t i o n system's c u r r e n t knowledge.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Knowledge Requirements of an Explanation Capability",
                "sec_num": null
            },
            {
                "text": "I f a n e x p l a n a t i o n c a p a b i l i t y also i s t a p ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Knowledge Requirements of an Explanation Capability",
                "sec_num": null
            },
            {
                "text": "The f i n a l t y p e o f knowledge t h a t some g e n e r a l q u e s t~o n -a n s w e r i n g t h e v a l u e is known w i t h c e r t a i n t y o r there are no r u l e s l e f t t o use. t h e n bo compare t h e answers t o these questions.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Other Domain-Independent Knowledge",
                "sec_num": "1.4.5"
            },
            {
                "text": "The ourpose of t h e explanation system is .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Scope of MYC~N'S Explanation Capability",
                "sec_num": null
            },
            {
                "text": "[preceded by the first 14 q u e s t i o n s in t h e c o n s u l t a t i o n ] . The f o l l o w i n g were used: ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Scope of MYC~N'S Explanation Capability",
                "sec_num": null
            },
            {
                "text": "[ 3.1 1 RULE027 n d i",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "15) Is",
                "sec_num": null
            },
            {
                "text": "WHAT ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "IS BLOOD A STERILE SITE?",
                "sec_num": null
            },
            {
                "text": "Each word in the dictionary has a synonym p o i n t e r t o i t s t e r m i n a l word (terminal words p o i n t t o themselbes). F o r the purpose o f analyzing t h e q u e s t i o n , a non-terminal word is considered t o be equivalent t o its ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Reducing the Question t o Terminal Words",
                "sec_num": null
            },
            {
                "text": "The 6 , 31, 351, 39, 4 1 , 42, 44, 347, 49 ",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 4,
                        "end": 42,
                        "text": "6 , 31, 351, 39, 4 1 , 42, 44, 347, 49",
                        "ref_id": "FIGREF3"
                    }
                ],
                "eq_spans": [],
                "section": "Determining What Pieces of Knowledge are Relevant",
                "sec_num": "2.4.3"
            },
            {
                "text": "The following r u l e s could have b?en used to determine t h a t t h e i d e n t i t y of ORGANISM4 is s t r e p t o c o c c u s : RULEO33. However, none of these succeeded i n t h e context o f ORGANISM-1. I f you would l i k e an e x p l a n a t i o n f o r why any of these rules failed, please enter their numbers:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "ORGANISM-I ?",
                "sec_num": null
            },
            {
                "text": "*# 33 Clause 2 of rule033 [\"the morpholoqy of the orqanasm i s coccusw]",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "ORGANISM-I ?",
                "sec_num": null
            },
            {
                "text": "was already known t o be false for ORGANISM-1, so t h e rule was never tried.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "ORGANISM-I ?",
                "sec_num": null
            },
            {
                "text": "It is definite t h a t the morphology of ORGAVISY-1 i s rod. mentioned t h e r e l e v a n t c o n t e x t .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "COCCUS?",
                "sec_num": null
            },
            {
                "text": "There are i n t e r a c t i o n s among t h e d i f f e r e n t types of questions which weren't mentioned e a r l i e r for t h e sake of c l a r i t y . Sometiaes a q u e s t i o n r a y i n d i c a t e a n incorrect assumption about the s a t e of MYCIN'S knowledge. For examole a q u e s t i o n could a s k \"Whv don't you think that ORGANISM-1 is E.coli?\" when, i n fact, the system has concluded t h a t the organism is E.coli. To answer t h i s q u e s t i o n , t h e e x p l a n a t i o n system would explain how i t ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "COCCUS?",
                "sec_num": null
            }
        ],
        "back_matter": [
            {
                "text": "The authors wish to express their qratitude for the inbrest and ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Acknowledgments",
                "sec_num": null
            }
        ],
        "bib_entries": {
            "BIBREF1": {
                "ref_id": "b1",
                "title": "To appear a s Machine R e p r e s e n t a t i o n s o f Knowledge p u b l i s h e d as Machine Intelligence",
                "authors": [
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Davis",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "King",
                        "suffix": ""
                    }
                ],
                "year": 1975,
                "venue": "",
                "volume": "8",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Davis, R., and King, J . An Overview of Production Svstems. To appear a s Machine R e p r e s e n t a t i o n s o f Knowledge p u b l i s h e d as Machine Intelligence 8 (eds. E.W. Elcock and D m Michie), John Wylie, December 1976. Also available as A 1 Memo 271, Stanford A r t i f i c i a l I n t e l l i g e n c e L a b o r a t o r y , S t a n f o r d University, October 1975.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Undersanding Naural Lanquaqe",
                "authors": [
                    {
                        "first": "T",
                        "middle": [],
                        "last": "Winoqrad",
                        "suffix": ""
                    }
                ],
                "year": 1972,
                "venue": "Coqnit ive Psycholosy",
                "volume": "3",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Winoqrad, T. Undersanding Naural Lanquaqe. Coqnit ive Psycholosy 3,)-191, 1972.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "text": "explanation c a p a b i l i t y n o t o n l y a d d s to t h e system s c r e d i b i l i t y , but also enables t h e non-expert user t o l e a r n from i t . Furthermore, clear e x p l a n a t i o n s a l l o w an e x p e r t t o cbeck the system's \" r e a s o n i n q f l , p o s s i b l y d i s c o v e r i n g the need f o r r e f i n e s e n t s and additions t o t h e system s knowledge b a s e u I n a d e v e l o p i n q system, a n e x p l a n a t i o n c a p a b i l i t v can be u s e d as a d e b u q~l n q aid t o verify that additions to t h e system a r e worckins a s t h e v should, T h l s paper d i x u s s e s t h e general c h a r a c t e r l s t l c s o f e x p l a n a t i o n systems. w h a t t y p e s of e x p l a n a b l o n s t h e v s h o u l d be a b l e t o g i v e , what t y p e s of nowl ledge w i l l b e needed i n o r d e r t o g i v e these e x p l a n a b i o h s , and how t h l s knowledge m l q h t be o r a a n l z e d . The e x~l a n a t i o n fae ~l l t v I n MYCIN [ 5 , 6 , 7 ] i s d i s c u s s e d as an 1 1 l u s t r a k i o n of how t h e various pnoblems miqht be approached.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF1": {
                "text": "e r s t a n d i n g The Q u e s t l o n . . . . . . . . . -2.5Answering the Q u e s t i o n. . . . . . . . 36 C o n c l u s i o n s . . . . . . . . . . I I .s u l t a t i v e P r o d u c t i o n Systems A c o n s u l t a t i o n proqram p l a y s t h e m l e o f an e x p e r t c o n s u l t a n t i n some domain, ~i v i n~ a d v i c e o r answers t o non-experts w i t h ~roblems i n t h e domain. Users w i l l o f t e n want t o know how t h e s v s t e a a r r i v e d a t i t s r e s u l t s d u r i n q a p a r t i c u l a r c o n s u l t a t i o n . T h i s Daper e x p l a l n s how t h e i m p l e m e n t a t i o n of such a proqram a s a ~r o d~c t i o n s y s t e m c a n f a c~l l t a t e",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF2": {
                "text": "p r o d u c t i o n system [2] consists of three basic components. a s e t of p r o d u c t i o n r u l e s , a d a t a base which is b o t h used and u p d a t e d bv these r u l e s , and a r u l e i n t e r p r e t e r . A p r o d u c t i o n rule o f t e n i s i n t h e form of a situation-action r u l e . i t describes a situation and a set o f a c t i o n s t o be taken i f t b i s s i t u a t i o n i s found t o exist. T h e r u l e i n t e r m e t e r d e t e r n i n e s t h e order I n which r u l e s w i l l be t r i e d , checks t o see i f t h e situations e x i s t , and u n d e r t a k e s the r e q u i r e d a c t i o n s . It a l s o d e t e r m i n e s how manv of t h e p o t e n t i a l l y u s e f u l r u l e s w j l l be used. o n l y t h e flrst (where o r d e r i n q may b e p r e d e t e r m i n e d or comouted d y n a m i c a l l y ) , all p o s s i b l e r u l e s , o r enough rules t o s a t i s f y some c r i t e r i o n that t h e i n t e r p r e t e r u s e s . I r . I n others, t h e o r d e r i n which r u l e s a r e t r i e d v a r i e s with d i f f e r e n t c o n s u l t a t i o n s , s i n c e a rule w i l l b e t r i e d as soon a s t h e r u l e i n t e r p r e t e r d e t e r m i n e s t h a t it may b e u s e f u l . I n s u c h s v s t e a s , t h e common a l t e r n a t i v e s are data-directed r u l e i n v o c a t i o n , i n which a r u l e i s considered \" u s e f u l f f i f i t s s i t u a t i o n part matches t h e data base, and qoal-d i r e c t e d r u l e i n v o c a t i o n , i n which a r u l e i s i f i t s a c t i o n p a r t will h e l p t h e system r e a c h i t s c u r r e n t g o a l . Many s y s t e m s u s e a combination of soaland d a t a -d i r e c t e d r u l e invocation. A c o n s u l t a t i v e p r o d u c t i o n system need n o t be a psycholoqical model, i m i t a t i n g a human s r e a s o n i n g p r o c e s s . The imporbant p o i n t is t h a t t h e System and a human e x p e r t u s e t h e same ( o r s i m i l a r ) knowledqe a b o u t t h e domain t o a r r i v e a t t h e same answer t o a g i v e n problem. The svstem's r u l e s and data base can be viewed as a knowledge base c o n t a i n i n q t h e domainspeclfic knowledge of an e x p e r t a s w e l l a s facts about a p a r t i c u l a r broblem. When a r u l e i s u s e d , i t s a c t l o n s make changes t o t h e d a t a b a s e which a r e t h e sy5tem's d e c i s i o n s o r d e d u c t i o n s . Thus, a r u l e can be t h o u q h t o f a s a plece of judgmental knowledqe, u s i n g the judqment and knowledge of au e x p e r t t o make d e d u c t i o n s . The p r o c e s s o f t r y i n g r u l e s and t a k i n g a c t l o n s can be t h o u q h t o f a s llreasonsngw, and e x p l a n a t i o n s c o n s i s t of showing how r u l e s used i n f o r m a t i o n p r o v i d e d by t h e u s e r t o make various i n t e r m e d i a t e d e~u c t i o n s and finally t o a r r i v e a t t h e answer. If t h e i n f o r m a t i o n c o n t a i n e d i n t h e s e r u l e s i s s u f f i c i e n t t o show why a n a c t i o n was taken ( w i t h o u t a e t t i n q i n t o programming d e t a i l s ) , an e x p l a n a t i o n can c o n s l s t o f p r i n t i n g each r u l e that was used ( o r an E n g l l s h e q u i v a l e n t o f what t h e r u l e means,",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF3": {
                "text": "System with Explanation CapabilityT h e three coniponents of a p r o d u c t i o n system ( a RULE INTERPRETER, a set of PRODUCTION RULES, and a DATA BASE 1 a r e augmented by a n EXPLANATION CAP4BILITY. Thed a t a b a s e i s made u p o f general facts a b o u t the system's domain of expertise, f a c t s t h a t the user e n t e r s about a specific problem, and d e d u c t i o n s made about t h e problem by t h e system's r u l e s . These deductions form the b a s i s of t h e s y s t e n ' s c o n s u l t a t i v e advice. The e x p l a n a t i o n c a p a b i l i t y makes use of t h e system's knowled_qe base t o q i v e the user explanations. This knowledae b a s e i s made U D o f s t a t i c domain-specif ic knowledge ( b o t h f a c t u a l and j u d m e n t a l ) and dynamf,~ knowledge specific to a particular problem' 1.2",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF4": {
                "text": "To ensure that the EC can handle questions about a l l relevant aspects o f the system'$ knowledge and actions. It should be capable of giving a few baslo types of explanations, f o r example.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF5": {
                "text": "i s t i n g u i s h two slighfly different functions f o r an EC and d i v i d e it into two components. the reasoning-status checker (RSC) t o be used during t h e consultation, and the general question answerqr (GQA) t o be used during the consultation or a f t e r the system h a s ~r i n t e d its results. A reasoning-status checker w i l l answer questions asked durinq-a consultation about the status of the systsb s reasoning process, A few s i m p l e commands are o f t e n s u f f i c i e n t t o h a n d l e t h e q u e s t i o n s t h a t t h e RSC is e x p e c t e d t o answen A g e n e r a question-answer w i l l answer q u e s t i o n s a b o u t t h e c u r r e n t s t a t e o f t h e syqtem s knowledge base, i n c l u d i n g b o t h s t a t i c domain khdwledge a n d facts accumulated d u a i n q t h e c o n s u l t a t i o n . A GQA will o f t e n need t h e a b i l i t y t o r e c o g n i z e a wide r a n g e of q u e s t i o n t y p e s about many a s p e c t s of t h e system's knowledge. For t h i s r e a s o n , i t m i g h t be d i f f i c u l t t o d e f i n e a few s i m p l e commands which would b e e a s y t o l e a r n and s t i l l c o v e r a l l t h e p o s s i b l e q u e s t i o n s t h a t might be asked. C o n s e q u e n t l y , n a t u r a llanguage p r o c e s s l n q i n t h i s component may be i m~o r t a n t t o an e x p l a n a t i o n system s a c c e~t a b i l i t y . I n a n i n t e r a c t w e consultatlor!, t h e svstem p e r i o d i c a l l y r e q u e s t s information a b o u t t h e problem. T h i s o f f e r s t h e u s e r a n o p p o r t u n i t y t o r e q u e s t e x p l a n a t i o n s w h i l e t h e c o n s u l t a t i o n i s i n p r o a r e s s . I n noni n t e r a c t i v e o o n s u l t a t l o n s , t h e u s e r h a s no o p p o r t u n d t v t o i n t e r a c t h i t h t h e system u n t l l a f t e r lt has p r i n t e d ~t s c o n c l u s i o n s . U n l e s s there 1s some mechanism a l l o w i n g a u s e r t o i n t e r r u p t t h e r e a s o n i n g p r o c e s s and a s k q u e s t i o n s , t h e e x p l a n a t i o n c a p a b i l i t y for such a system w i l l be l i m i t e d t o q u e s t i o n s a b o u t t h e system's f i n a l knowledge s t a t e . It w i l l have no",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF6": {
                "text": "is organized. In order t o give e x p l a n a t i o n s o f t h e svstem's c u r r e n t ( o r p r e v i o u s ) actions, an EC also needs t o understand how t h e system s r u l e i n t e r p r e t e r works. when ruces w~l l b e tried how t h e y can fall, what c a u s e s t h e i n t e r p r e t e r t o try one r u l e b u t not a n o t h e r etc. This q e n e r a l \"schemaw f o r how o r why c e r t a i n r u l e s are used, t o q e t h e r with a comprehensive record of t h e specific a c t i o n s taken d u r i n g a p a r t i c u l a r c o n s u l t a t i o n , can be used a$ a basis f b r e x p l a i n i n g t h e r e s u l t s of that c a n s u l t a t i o n . A r e a s o n i n g -s t a t u s checker w i l l need d record o f what t h e system h a s done s o f a r i n o r d e r t o e x p l a i n how i t arr4ved a t t h e c u r r e n t s t e~. General knowledge bf how t h e rule i n t e r p r e t e r works i s necessary i n o r d e r t o e x p l a i n where t h e c u r r e n t s t e p w l l l lead. The a b l l i t y t o understand individual",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF7": {
                "text": "r o v i d e i n f o r m a t i o n about how t h e s y s t e m a r r i v e d a t t h e f a c t s t h a t are c u r r e n t l y i n i t s dynamic knowledge b a s e , t h e GQA w i l l need a l l t h e i n f o r m a t i o n t h a t a r e a s o n i n ps t a t u s c h e c k e r u s e s . a d e t a i l e d r e c o r d o f t h e c o n s u l t a t i o n , a n understand in^ o f the r u l e s n t e r p r e t e r , and the a b i l i t y t o u n d e r s t a n d r u l e s . T h e s e t h r e e t y p e s of knowledge c o u l d be supplemented w i t h a limited amount o f q e n e r a l i n f o r m a t i o n a b o u t s u c h t h i n q s a s e l e m e n t a r y l o s i c , s e t t h e o r y , and a r i t h m e t i c comparisons. T h i s would a l l o w t h e GQA to answer more c o m p l i c a t e d q u e s t i o n s a b o u t why t h e s y s t e m ' s knowledge base IS i n b i t s c u r r e n t s t a t e , and t o answer q u e s t i o n s i n v o l v i n g p e l a t i o n s h i~s between different f a c t s i n t h e knowledge base. The n a t u r e o f t h e c o n s u l t ' a t i o n domain as w e l l a s what primarv purpose t h e e x p l a n a t i o n c a p a b i l i t y is t o s e r v e wlll i n f l u e n c e t h e r a n g e o f questions t h a t a n EC s h o u l d h a n d l e . I n some s y s t e m s , a s l m~l e retrieval o f f a c t s may s u f f i c e , w h i l e others may need t o g i v e d e t a i l e d d e s c r i p t i o n 01 t h e p r o d u c t i o n s y s t e m ' s l t d e c l s i o n f l D r o c e s s and t o make a number o f deductions from f a c t s t h a t i t has. Knolwledge Requirements of an Explanation Capability Access to t h e c o n s u l t a t i o n system's knowledge base is a prerequisite for performance of t h e explanation c a p a b i l l t v . Other t y p e s of knowledqe may be added to t h e system to e n a b l e the EC t o answer a wider range of q u e s t l o n a , Deaign C o n s i d e r a t i o n s The l a s t two s e c t i o n s d e s c r i b e d what a n e x p l a n a t i o n c a p a b i l i t y i s , b u t l i n i n q what tasks i t should perfom, a n d what i t requires i n order t o p e r 1 om these t a s k s , I n t h i s l s e c t i o n , w e d i s c u s s d e s i g n c o n s i d e r a t i o n s f o r t h e p a r e n t p r o d u c t i o n system t h a t w i l l e n a b l e i t s EC t o meet t h e r e q u i r e m e n t s t h a t were o u t l i n e d i n t h e p r e v i o u s s e c t i o n . This d i s c u s s i o n is n o t meant t o d e f i n e t h e v c o r r e c t w way of r e p r e s e n t i n g o r o r g a n i z i n g knowledge, b u t r a t h e r t o mention c e r t a i n f a c t o r s which s h o u l d be t a k e n i n t o a c c o u n t when d e c l d i n g what r e p r e s e n t a t i o n o r o r q a n i z a t i o n w i l l be best f o r a g i v e n product I o n system. 1.4.1",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF8": {
                "text": "t e p ~s t o d e c i d e what basic types o f questions the system s h o u l d b e able t o answer. This w i l l have a d i r e c t I n f l u e n c e on how t h e EC is i m~l e m e n t e d . It i s i m p o r t a n t , however t o make t h e r n i t i a l d e s l~n f l e x i b l e enough t o accomodate possible f u t u r e a d d i t i o n s t o t h e set o f basics. I f t h e b a s i c forms are d i v e r s e enough, some l e v e l o f n a t u r a llanguaqe u n d e r s t a n d i n g may be n e c e s s a r y . The degree of s o p h l s t l c a t i o n o f the natural-languacze p r o c e s s o r w i l l depend w o n what kind o f ~e r f o r m a n c e is.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF9": {
                "text": "r g a n i z a t i o n of Knowledge The format and o r g a n i g a t i o n o f v a r i o u s components of the p r o d u c t i o n system's knowledge b a s e w i l l affect t h e d e s l g n of an EC. I n d i v i d u a l ~i e c e s o f s t a t i c and dynamic knowledge presumably w i l l b e o r g a n i z e d m some f a s h i o n which makes them accessible d u r i n g t h e c o n s u l t a t i o n . A GOA f a c i l i t y c o u l d make u s e o f such o r g a n i z a t i o n t o h e l p I n f i n d i n g t h e i n f o r m a t i o n needed bo answer a q u e s t i o n . The l e s s o r g a n i z e d t h e knowledge b a s e t h e more d i f f i c u l t wxlf, b e t h e t a s k of t h e EC, as more c o m p l i c a t e d r o u t i n e s must be u s e d i n o r d e r t o f i n d t h e d e s i r e d ~n f o r m a t i o n . During t h e cQurse o f t h e c o n s u l t a t i o n , t h e system s h o u l d keep a r e c o r d o f its a c t l o n s f o r u s e by b o t h Co~nponents o f t h e e x p l a n a t i o n c a p a b i l i t y . Where t h e o r d e r i n g o f e v e n t s is i m p o r t a n t ( e .~; . when t h e a c t i o n o f one r u l e e s t a b l i s h e s t h e s i t u a t i o n n e c e s s a r y f o r a s u b s e q u e n t r u l e t o yuuceed), t h e r e c o r d s h o u l d b e s t r u c t u r e d i n a madner which r e f l e c t s t h e o r d e r i n g o f e v e n t s as well a s t h e r e a s o n s whv e a c h e v e n t o c c u r r e d . 1.4.3 Knowledge o f What R u l e s Mean The e x p l a n a t i o n capability w i l l need t o u n d e r s t a n d some of t h e s e m a n t i c s o f i n d l v l d u a l production r u l e s . T h l s r e q u i r e m e n t c o u l d b e met by h a v i n g t h e svstem's knowledge base i n c l u d e a d e s c r i p t i o n o f what e a c h P u l e means, encoded i n some form which would b e o f u s e t o the EC. I f t h e format o f t h e system's r u l e s i s h i g h l y s t y l i z e d and w e l l -d e f i n e d , however, i t might be p o s s i b l e i n s t e a d t o implement a mechanism f o r ' I r e a d~n g \" the r u l e s . t h e language i n which t h e r u l e s t h e m s e l v e s are w r i t t e n c o u l d b e d e f i n e d . A h i g h -l e v e l d e s c r i p t i o n o f t h e i n d i v i d u a l components o f t h i s l a n g u a g e , t e l l i n g what each component means, c o u l d b e u s e d t o e n a b l e t h e EC t o read and u n d e r s t a n d r u l e s . If t h e r u l e s e t c o n s i s t s of a large nuaber o f r u l e s , and t h e s e r u l e s are composed e n t i r e l y o f a r e l a t i v e l y small number of p r i m i t i v e e l e m e n t s , t h i s second approach h a s t h e a d v a n t a g e t h a t l e s s i n f o r m a t i o n n e e d s t s be s t o r e d -a d e s c r i p t i o n o f e a c h o f t h e o r i m i t~v e components, as opposed t o a d e s c r i p t i o n of e a c h r u l e . When new r u l e s a r e added t o t h e system, t h e f i r s t approach r e q u i r e s t h a t d e s c r i p t i o n s of these r u l e s must be added. With t h e second a p p r o a c h , p r o v i d e d t h a t t h e new r u l e s are made up of t h e s t a n d a r d r u l e components, no a d d i t i o n a l d e s c r i p t i t e i n f o r m a t i o n would b e needed by t h e e x p l a n a t i o n c a p a b i l i t v . 1.4.4 Knowledge of t h e Rule I n t e r p r e t e r E n a b l i n g a n EC t o u n d e r s t a n d how t,he r u l e i n t e r p r e t e r works i s a n a l o q o u s t o e n a b l i n g l t t o u n d e r s t a n d r u l e s I t must be a b l e t o \"read\" t h e i n t e r p r e t e r o r e l s e it must have access t o some s t o r e d d e s c r i p t i o n o f how t h e i n t e r p r e t e r works. ?here i s a thirkl a p p r o a c h f o r u n d e r s t a n d l n q t h e rule i n t e r p r e t e r , one whlch would n o t be feasible for understand in,^ a l a r s e number o f r u l e s . Knowledge o f how t h e i n t e r p r e t e r works c o u l d b e b u i l t into t h e EC -t h e information would n o t be s t a t e d explicitly, b u t would be u s e d implicitly by t h e programmer I n w r i t i n g t h e a c t u a l code f o r t h e e x p l s n a t i o n capability. The EC can b e t h o u s h t o f a s a number o f m s~e c 1 a 1 i s t 3 f f , each c a p a b l e o f g l v i n g a s i n g l e t y p e o f e x p l a n a t i o n . T h e r e could he one s p e c i a l i s t f o r e a c h o f t h e b a s l c q u e s t l o n t v~e s t h a t t h e s y s t e n can answer. Each of t h e s p e c i a l i s t s n e e d s o n l y a small amount of ~n f o m a t i o n about t h e r u l e i n t e r p r e t e r whlch c o u l d be b u i l t I n t o ~t s l t e x p l a l n l n~n p r o c r e a .",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF10": {
                "text": "f a c i l i t i e s w l l l need i s l n f o r m a t l o n a l l o w i n q d e d u c t i o n s t o be vade froa f a c t s I n t h e knowledge b a s e . The r e~r e s e n t a t i o n and e x t e n t oC t h l s knowledge w i l l depend upon t h e t y p e s of questions t h a t t h e system is t o answer I f logic is needetl onlv t o d e t e~m i n e t h e answers t o questions of a c e r t a i n tvpe f o r example, t h e necesaarv deductions could b e b u i l t i n t o the s p e c i a l i s t f o r answerinq t h a t t y p e of question. On t h e o t h e r hand, i n aowe explanation capibilities, the GQA will b e expanded t o do more than simply q i v e explanations o f t h e system s a c t i o n s or t o q u e r y its data base -i t will be expected t o answer a wide range of questions involvlnq various kinds sf inferences about t h e knowled~e base. Such a CQA w i l l need t o check f o r e q u a l i t y o r set m e m b e r s h i p , make a r i t h q e t i c compariSons, o r make l o g i c a l deductions. I n general most information of t h i s t y p e can b e embodied In a new kind of s p e c i a l i s t which is an e x p e r t a t some s o r t o f l o g i c a l deduction o r comparison. Representation of this s o r t of qeneral knowled~e w i l l become important a s t h e GO4 becomes not s i n l p l y an explanation t o o l , b u t a l s o a deduct i v e one. is an example o f a production-based c o n s u l t a t i o n system with a well-developed e x p l a n a t i o n c a p a b i l i t y . A p r o d u c t i o n r u n i s an i n f e c t i o u s disease t h e r a p y c o n s u l t a t i o n i n which MYCIN i s t h e i n f e c t i o u s dlseaae e x p e r t , and t h e user is a d o c t o r who wants a d v i c e a b o u t t h e t r e a t m e n t of a p a t i e n t . Knowledge t h a t i s gathered d u r i n g t h e c o n s u l t a t i o n i s organized i n t o a t t r i b u t e -o b j e c t -v a l u e t r i p l e s . In response t o q u e s t i o n s d u r i n g the c o n s u l t a t i o n , t h e u s e r e n t e r s i n f o r m a t i o n a b o u t t h e e x i s t e n c e o f s e v e r a l o b j e c t s , called contexts. t h e p a t i e n t , i n f e c t i o n s t h a t the p a t i e n t has, organisms which may be causing t h e s e i n f e c t i o n s , c u l t u r e s t h a t were t a k e n , and d r u g s t h a t were q i v e n , The t a s k of t h e c o n s u l t a t i o n system i s t o d e t e r m i n e t h e v a l u e s o f v a r i o u s a t t r i b u t e s ( c a l l e d clinical ~a r a m e t e r s ) of t h e s e c o n t e x t s . For example, AGE is a c l i n i c a l parameter of t h e p a t i e n t ; IDENTITY i s a c l i n i c a l p a r a m e t e r o f an o r e a n i s m , w i t h STRFPTOCOCCUS a s a possible v a l u e ; SITE i s a parameter o f a c u l t u r e , with BLOOD as a p o s s i b l e v a l u e . A c l i n i c a l p a r a m e t e r ' s v a l u e may be d e t e r m i n e d by a s k i n g t h e user, o r by using d e c i s i o n r u l e s . The parameter is s a i d t o b e rvtracedw when t h e syBtem h a s done a l l i t can t o f i n d o u t the parameter's v a l u e . T r a c i n q a parameter i n v o l v e s a s k i n g t h e u s e r f o r a v a l u e (where a p p l i c a b l e ) and trying r u l e s f o r d e t e r m i n i n g t h e v a l u e of that p a r a m e t e r . R u l e s are t r i e d u n t i l",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF11": {
                "text": "i s i o n r u l e has a s i t u a t i o r l p a r t called i t s PREMISE. This c o n 9 i s t s o f p r e d i c a t e s , c o n d i t i o n s that are t e s t e d t o determine whether the i n d i c a t e d s i t u a t i o n e x i s t s . I f . t h e c o n d i t i o n s i n a r u l e ' s PREMISE are t r u e , its ACTION w i l l be e v a l u a t e d , giving new (or updated) values t o some parameter(s) . Before a a o n d r t i o n i n a r u l e ' s PREMISE can be t e s t e d , t h eparameters that it mentions must be traced.For example, before r u l e 209 (below) can succeed, the system must know t h e s i t e of the c u l t u r e , t h e p o r t a l of e n t r y o f t h e organism, and whether the p a t i e n t is a compromisedhost. If any of t h e clauses in the PREMISE is f a l s e , o r if t h e system is unable t o find out t h e value of one of t h e s e parameters, t h e rule will fail. ----111 (PREMISE) I f : 1 ) T@e site o f t h e c u l t u r e i s b l o o d , and The p o r t a l of e n t r y of t h e organism i s GI, and 3 ) The p a t i e n t i s a compromised h o s t (ACTION) Then: It 1s d e f i n l t e ( 1 -0) t h a t b a c t e r o i d e s i s an organism for which t h e r a p y should cover Associated w i t h each a t t r i b u t e -o b j e c t -v a l u e t r i p l e i s a c e r t a i n t y factor -a number between -1 and 1 i n c l u s i v e which indicates how s t r o n q l y as a t t r i b u t e , object, and value. Each c o n t e x t is named uniquely, allowing the system t o refer t o meaning t h e second culture, o r ORGANISM-3, meaning t h e t h i r d organism. Moreover, the c o n t e x t s are orqanized i n t o a tree known a s t h e context tree, which defines r e l a t i o n s h i p s among them. For example, an organism is t h e d i r e c t descendent of' t h e c u l t u r e from w h i c l~ it was isolated. In the portion of a tree shown in FQure 3 ORGANISM-3 hangs under CULTUR'E-2 i n d i c a t i n q t h a t STREPTOCOCCUS was i s o l a t e d from t h e BLOOD c u l t u r e . Portion of a Context Tree Showing Some Contexts, Clinical Parameters, and Values The rule i n t e r p r e t e r (MYCIV'S c o n t r o l s t r u c t u r e , described i n detail i n ['TI) c h o a s e s t h e pules which should be used i n t h e particular c o n s u l t a t i o n , i n t e r p r e t s t h e s e r u l e s , and creates a r e c o r d o f ~t s a c t i o n s f o r use by t h e e x p l a n a t i o n system. Rules are invoked t o find out v a l u e s of parameters i n a given c o n t e x t . A r u l e i s a p p l i e d t o t h e lowest c o n t e x t i n t h e c o n t e x t tree whose parameters are mentioned by t h e r u l e . The r u l e can use ( o r conclude a b o u t ) parameters of t h i s c o n t e x t , o r of anv c o n t e x t which 1s i t s ancestor I n t h e tree. For example, i f RULE209 were apolied t o ORGANISM-3 (see Figure 3) i t would need t h e S I T E of the culture from which t h e STREPTOCOCCUS was i s o l a t e d . The tree i n d i c a t e s t h a t t h i s is CULTURE-2, Rather t h a n being a s e q u e n t i a l c y c l e throuqh t h e rule s e t , wher-e e a c h rule i s t r i e d i n some predetepmined order, t h e flow o f c o n t r o l is qoaldirected. This means that only rules which conclllde about t h e c u r r e n t goal (to f i n d out the value of a given parameter) are examined. The PREMISE of one of these rules may need t o u s e some ~a r a m e t e r whose v a l u e is unknown, This s e t s up a subgoal, namely t o deterqine t h e v a l u e of t h i s parameter s o that t h e r u l e can be used. MYCIN's goal-direcrted approach means t h a t t h e system (and n o t the user) takes t h e initiative d u r i n g a c o n s u l t a t i o n . The user w i l l be asked about onLy t h o s e parameters Qhich way be relevant t o t h e p a r t i c u l a r p a t i e n t ' s case. 2.OrganizatYon of Knowledge i n MYCXN I n o r d e r t o give e x p l a n a t i o n s of a c o n s u l t a t i o n system's d e c i s i o n s ,an e x p l a n a t i o n c a p a b i l i t y must have a c c e s s t o t h e s y s t e m ' s knowledge base.More informative e m l a n a t i o n s can be g i v e n i f t h e EC also has knowledae ofhow t h e syfitem works, a record of t h e c o n s u l t a t i o n , and possibly some domain-independent knowledae. This s e c t i o n discusses how MYCIN meets these requirements. The system's knowledee base consf sts o f s t a t i c medical knowledge p l u s dynamic knowled3e about a s p e c i f i c c o n s u l t a t i o n . S t a t i c knowledqe is f u r t h e r classified as factual and judgmental. F a c t u a l k n o w l e d~e c o n s i s t s of facts which are medically v a l i d i n d e~e n d e n t of t h e ~a y t i c u l a r case. Judgemental knowledge c o n s i s t s o f product ion r u l e s r e p r e s e n t i n g deduct ions which might be made, c o n d i t i o n a l an what i s a l r e a d y known about t h e case. The format o f production r u l e s and o f dynamic knowledge h a s a l r e a d y been d e s c r i b e d . 2.2.3 Organization of Factual Knowledge As discussed i n Section 2 . 1 , a l l knowledge which is gathered durinq t h e consultation i s organized into attribute-object-value triples. For consistency, many facts i n t h e static knowledqe base also have t h i s format. This includes objects such as bacteria and antibiotics, and attributes such a s t h e staininrg characteristics of a bacterium or t h e recommended dosage of an antibiotic : ATTRIBUTE OBJECT VALUE ------. I -------------GRAM E .COLI GRAMNEG DOSE GENTAMICIN 1.7 mq kg q8h IV (or IM) The remainder of the factual knowledge consists o f l i s t s and tables: pieces of aedical knowledge, organized m such a way t h a t they can be used t o augment t h e producbion rules. For example, one such piece of knowledqe is the l i s t of the possible culture s i t e s which are normallv nonsterile, NONSTERILESITES : ( CERVIX CUTANEOUS-ULCER LOCHIA NOSE SKIN STOOL THROAT URETHRA VAGINA) The likely pathogens associated with the different culture s i t e s are organized i n a table, with different entries for the different s i t e s , PATH-FLORA -------..--THROAT : (STREPTOCOCCUS-PNEUMONIAE STREPTOCOCCUS-GROUP-A NEISSERI A-MENINGITIDIS ) URINE: (E.COLI PSEUDOMONAS ENTEROCOCCUS PROTEUS KLEBSIELLA ENTEROBACTER) SKIN : (STAPHYLOCOCCUS-COAG-POS STREPTOCOCCUS-GROUP-A STAPHYLOCOCCUS-COAG-NEG) CERVIX: ( STREPTOCOCCUS CLOST RIDIUM-GANGRENE NEISSERIA-GONORRHA STREPTOCOCCUS-GROUP-A ) Production rules can make use of t h i s tabularized information: RULE058 --\"-I--If: 1 ) The s i t e of the culture i s one o f : those sites that are normally nonsterile, and 2 ) This organism and at least one of t h e likely pathogens associated with t h e s i t e of t h e c u l t u r e agree wi'th respect t o t h e following properties: gram morph air Then: There i s strongly sugaestive evidence (.9) that each o f these pathogens is the i d e n t i t v of the organism Note that the information i n t h e table could have been 0r~anized as attribute-object-value triples (where the object would be a culture site). If t h i s had been done, however, the above rule could not have been written. To accomplish the same purpose (without a change i n t h e control structure), the svstem would have needed several rules -a separate one for each entry i n the table. Structurinq certain facts into l i s t s an.d tables er~ables individual production rules t o exDress general theories which allow a number of specific deduct ions t o be made. MYCIN'S approximately 400 rules i s composed of a small number of conceptual primitives. A t o t a l of 60 such primitives make up the language i n which rules &re written, This design f a c i l i t a t e d the implementation of a mechanisp for translating rules i n t o Enqliah ( d e s c r i b e d i n detail i n 171). Each p r i m i t i v s functions has a translation template w i t h blanks t o be filled i n with translations of the function's arquments. A larrr;e part of MYCIN'S exolanation c a p a b i l i t v depends on t h i s a b i l i t y to translate rules into a form t h a t the user can understand. Having a small number of rule components a l s o f a c i l i t a t e s the examination of rules t o see which might be apnlicable t o t h e explanation a t hand. MYCIN'S knowledge of production r u l e s , therefore, takes t h e form o f a general mechanlsrn for \"readinq\" rules, On t h e o t h e r hand, no attempt has been made t o read t h e code o f the rule interpreter. Procedural knowledge about the interpreter is embodied i n n s p e c i a l i s t s u , epch capable of answerinq a single type o f question. Each specialist knows how the r e l e v a n t part of the c o n t r o l structure works and what p i e c e s of knowledge i t uses. I n order t o understand r u l e s , the system's various specialists use a small amount o f knowledge about rules in general, tosether with descriptions or templates of each o f the rule components. A s an example, t h e following rule i s composed o f t h e units SAND, SAME, and CONCLUDE. RULE009 -------PREMISE: (SAND (SAME CNTXT GRAM GRAMNEG) (SAME CNTXT MORPH COCCUS)) ACTION : ( CONCLUDE CNTXT IDQNTITY NEISSERIA TALLY 800) [Translation: If: 1 ) The gram s t a i n of the o r~a n i s m i s qrarnneg, and 2) The morphology of the organism i s coccus Then: There i s strongly suggestive (-8) that the identity of the orqanism i s Neisseria] [When the rule i s u s e d , the LISP atom CNTXT is bound t o some object, the context t o which the rule i s a p~l i e d (see Section 2.111 The template for CONCLUDE i s shown below. This describes each o f t h e arguments t o t h e f u n c t i o n : f i r s t , an object ( c o n t e x t ) ; second, an attribute ( c l i n i c a l parameter); t h i r d , a value for this parameter; fourth, the t a l l y or degree of c e r t a i n t y of t h e P,REMISE; and l a s t , t h e c e r t a i n t y factor -a measure o f how strong our belief i n t h i s conclusion would b e , assuming bhat the PREMISE of the rule i s d e f i n i t e l y true. CONCLUDE I-------TEMPLATE: (CNTXT FARM VALU TALLY CF) To i l l u s t r a t e how t h i s is used, consider an e x p l a n a t i o n t h a t i n v o l v e s f i n d i n g all r u l e s which c o u l d cbnclude t h a t t h e i d e n t i t y of an organism i s Neisseria. The appropriate swe.cialist would s t a r t with those r u l e s which t h e avstem u s e s t o conclude v a l u e s f o r the parameter IDENTITY. Using t e m p l a t e s o f t h e v a r l o u s ACTION f u n c t i o n s which appear i n each o f these rules, t h e specialist picks out onLy t h o s e (like RULE009) which have NEISSERIA in t h e i r V A L J s l o t . ThQ also i l l u s t r a t e s t h e sort o f knowledqe that can b e b u i l t i n t o a specialist . The s p e c i a l i s t knew t h a t t h e c o n t r o l s t r u c t u r e uses s t o r e d l i s t s t e l l i n g which r u l e s can be used t o determine t h e v a l u e of each parameter. Furthermore, ~t knew t h a t l t was n e c e s s a r y t o look o n l y a t t h e r u l e s ' ACTIONS because it 1 s t h e ACTION t h a t concludes f a c t s , while the PREMISE uses f a c t s . of t h e e x p l a n a t i o n capability's specialists need a record o f the c o n s u l t a t i o n . This record 1s b u i l t d u r i n g t h e c o n s u l t a t i o n , and is organized into a t r e e s t r u c t u r e called t h e history tree which reflects MYCIN'S goal-directed approach. Each node i n the tree r e p r e s e n t s a goal and c o n t a i n s Information about how t h e system tried t o accomplish this g o a l : bv a s k i n g t h e user o r by t r y i n g r u l e s . Associated with each r u l e i s a r e c o r d of whether t h e rule succeeded, and i f n o t , why it f a i l e d . If t r y i n g some r u l e c a u s e s t h e system t o trace a new parameter, thereby setting up a subgoal, the node for t h i s subgoal is the offspring of the node containins the rule which caused the t r a c i n g . Figure 4 i l l u s t r a t e s how part of ii h i s t o r y tree miwht look, In this example, R1JLE003 caused tracing of t h e parlameter CATEGORY which i s used i n the PREYISE of t h i s rule, Portion of a History Tree [ R U L E O O~ is shown above, see Figure 5 for RULE003 and RULE0373 goal: IRENTITY of ORGMISM-1 ask: question 7 rules: RULE009 ( f a i l e d , clause 1 ) ... RULE003 (succeeded) CATEGORY of ORGANISM-1 rules: RULE037 (succeeded) ...",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF12": {
                "text": "wasn 't ade, t h e appropriate specialist uses t h e logical conclusion t h at the answer consists o f e x~l a i n i n g what prevented the system from u s i n g each o f [no rules] the rules t h a t would have made t h a t decision.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF13": {
                "text": "If d e d u c t i o n s or comparisons are needed to answer questions of a specific t y p e , t h e n t h e necessary logic is b u i l t into the appropriate specialist. There i s no general r e p r e s e n t a t i o n of knowledge about logic, arithmetic, o r set t h e o r y that t h e explanation capability can use t o make inferences from d i f f e r e n t facts in its knowledge base. To find out whether ORGANISM-I and ORGANISM-2 have t h e same identity, f o r example, it is necessary f o r t h e user t o ask separately f o r t h e i d e n t i t y of each organism,",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF14": {
                "text": "t o enable a user t o see how MCIN makes decisions, both i n general and with r e f e r e n c e t o a particular consultation. To make t h i s f a c i l i t y as useful as possible, we have t r l e d t o a n t i d p a t e a11 t y p e s of q u e s t i o n s whlch a user rnlpht ask, and t o make every part of t h e system's khowledge base and reaqoning process accessible through clear explanations, The entire explanation f a c i l i t y consists of a number of components or wspecialistsm each capable of giving a single type o f explanation. These components are grouped into three sets: one for explaining what the system i s doing a t a given time, one for answering questions about t h e system's s t a t i c knowledge base, and one f o r answering quest i o n s about t h e dynamic knowledge base. The f i r & s e t forms MYCIN 's reasoning-status checker ; the second and t h i r d together make up the system s general question-answer.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF15": {
                "text": "o f MYCIN'S r e a n o n i n q c h a i n bv a s k i n g WHY this piece of information i s important. As explained i n s e c t i o n s 2. 1 and 2.2, the system asks a question i n order to f i n d out about i t s curpent goal. Conslder t h e p o r t i o n of a history tree shown i n F i g u r e 4 . HOSPITAL-ACQUTRED i s one subgoal, CATEGORY 13 a n o t h e r a t t h e next level up, atid RULE037 links them, The ffreasonff f o r adkinq whether t h e infection was hospital-acquired, then, is based on an attempt t o use t h i s rule t o determxne t h e h i~h e r subgoal. a WHY, t h e user may a s k HOW this qoal was (or w i l l b e ) a c h i e v e d . MYCIN ' s reasoning-status checker i s described i n more d e t a i l i n [3,4,61",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF16": {
                "text": "b",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF17": {
                "text": "t h e p a t i e n t ' s i l l n e s s w i t h ORGANISM-1 a h o s p i t a l -a c q u i r e d i n f e c t i o n ? ** WHY [i.e. WHY i s i t important t o determine whether o r n o t t h e i n f e c t i o n with ORGANISM-1 was a c q u i r e d while t h e p a t i e n t was hospitalized? ] f l .Ol This will a i d i n d e t e r m i n i n g t h e c a t e g o r y of ORGANISM-!. It ha3 a l r e a d y been established t h a t [1.1] t h e gram s t a i n o f ORGANISM-1 i s gramneq, and [ I .2] t h e morphology of ORGANISM-1 i s rod, and [1.3] t h e a e r o b i c i t y of ORGANISM-1 i s f a c u l t a t i v e T h e r e f o r e , i f [ 1.41 the ineection with ORGANISM-1 was n o t a c q u i r e d while the p a t i e n t was h o , s~i t a l ized t h e n there i s s t r o n g l y suggestive e v i d e n c e ( 8 ) that t h e c a t e a o r y o f ORGANISM-1 i s e n t e r o b a c t e r i a c e a e ALSO: there i s weakly dugqeetive e v i d e n c e (. 1 ) t h a t t h e i d e n t i t y of ORGANISM-1 i s pseudomonas-aeruginosa 1 [RULE0371 WHY [i.e. WHY i s it i m p o r t a n t t o d e t e r m i n e t h e c a t e g o r y of ORGANISM-I?] [2.0]. . . i n brder t o d e t e r m i n e t h e identity of ORGANISM-1 It has already been e s t a b l i s h e d t h a t C2.11 this b l w d c u l t u r e was taken from a sterile s o u r c e Therefore, i f [2.2] t h i s c u r r e n t organism and a t l e a s t one of t h e l i s t of members associated with t h e c a t e g o r y of the orqanism agree with r e s p e c t t o t h e following p r o p e r t i e s : a i r conformation then There i s s t r o n g l y suggestive evidence (.9) that each o f them is t h e i d e n t i t y of ORGANISM-1 [RULE003 1 ** HOW 1.3 [i.e. HOW was i t established t h a t tke a e r o b i c i t v o f ORGANISM-1 i s f a c u l t a t i v e ? 1",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF18": {
                "text": "Since t h i s gave a cumulative CF of ( ' 8 ) f o r f a c u l t a t i v e , and (.2) for a n a e r o b i c , i t has been e s t a b l i s h e d t h a t t h e a e r o b i c i t y of ORGANISM-1 i s facultative ** HOW 1.1 [i.e. HOW was it established that t h e gram stain of ORGANISM-1 i s qramneg?] You s a i d s o [ q u e s t i o n 1 1 3 .",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF19": {
                "text": "MPCIN's Reasoning-Status Checker [ u s e r e n t r i e s follow t h e double asterisks] 2.3.2MYCIN'a General Question AnswererThe q u e s t i o n -a n s w e r i n g p a r t of the systerll h a s natural-lanqusqeroutines f o r analyzinq t h e user's i n p u t . The syatem r e c o a n i z e s questions p h r a s e d i n a number of ways, thereby makinq the q~e s t~i o n -a n s w e r i n g f a c i l -l t v easier to use. Questions about t h e s t a t i c knowledge base may deal w i t h judgmental knowledge ( e . g . , which rules use or conclude a c e r t a i n piece of i n f o r m a t i o n ) o r they may ask about f a c t u a l knowledge -e n t r i e s in t a b l e s and lists. Some questions about s t a t i c knowledge are shown i n F i g u r e 6.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF20": {
                "text": "Sample Q u e s t i o n s a b o u t MYCIN'S Static Knowledge Perhaps t h e more i m p o r t a n t art of t h e question-answering svstem i s its a b l l i t y t o answer q u e s t i o n s about a articular c o n s u l t a t i o n . While some u s e r s may be i n t e r e s t e d In checklng the e x t e n t o f MYCIN'S s t a t i c knowledge, most questions w i l l ask for a j u s t i f i c a t i o n o f , o r f o r t h e r a t i o n a l e b e h i n d , p a r t i c u l a r decisf on3 which were made d u r i n g the c o n s u l t a t i o n . O u t l i n e d i n F l g u r e 7 are t h e t y p e s of q u e s t i o n s about dynamlc knowledqe which can b* handled at present. A few examples of each type are given. <Cntxt> i n d i c a t e s some c o n t e x t which was d i s c u s s e d in t h e ~n n s u l t a t l o n ; <parm> i s some clinical parameter of t h i s c o n t e x t ; <rule> i s one of the system's decision r u l e s .",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF21": {
                "text": "Sample Questions about a ConsultationBefore a question can be answered, it must be classif$ed a3 belonginq t o one ,of these groups. AsFigure 7illustrates, each question type includes a variety of ways in which the question can be worded, some specifying the parameter's value, some phrased in t h e negative, and soforth. MYCIN'S natural-language Drocessor must classify the questions, then determine what cllnLcal parameters, etc. t h e question references.The main emphasis i n t h e development of t h e MYCIV system h a s been the c r e a t i o n of a production system which can provide sound d i a q n o a t i c and t h e r a p e u t i c a d v i c e i n the field of i n f e c t i o u s disease. The explanation sfatem was included i n t h e system's o r i g i n a l d e s i q n i n order t o make t h e c o n s u l t a t i o n program's d e c i s i o n s a c c e p t a b l e , j u s t i f i a b l e , and i n s t r u c t i v e . S i n c e t h e question-answerina f a c i l i t y was n o t t h e primary focus o f t h e research, it is not designed t o b e a sophisticated natural-lanquage u n d e r s t a n d e r . Rather, i t u s e s c r u d e t e c h n i q y e s , r e l y i n g s t r o n g l y on t h e v e r y s p e c i f i c vocabulary of t h e domain, t o l l~n d e r s t a n d~~ whst i n f o r m a t i o n i s being requested. The analysis o f a q u e s t i o n i s broken i n t o three phases: t h e first c r e a t e s a list of terminal or r o o t words; t h e second d e t e r m i n e s what type of q u e s t i o n i s beinq asked (see t h e c l a s s i f i c a t i o n o f q u e s t i o n s i n Section 2.3); and t h e l a s t d e t e r m i n e s what p a r t i c u l a r parameters, lists, etc. are r e l e v a n t t o the q u e s t i o n . I n t h e first and l a s t s t e p s , t h e system d i c t i o n a r y is i m p o r t a n t , The d i c t i o n a r y Contains approximately 1400 words t h a t are commonly used in the domain of i n f e c t i o u s disease. It ificludes a l l words that are a c c e p t a b l e v a l u e s f o r a parameter, common synonyms of t h e s e words, and words used elsewhere by t h e system i n d e s c r i b i n g the parameter ( e . q . , when translating a rule i n t o E n g l i s h o r r e q u e s t i n g t h e value of t h e parameter).",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF22": {
                "text": "terminal) aynonym. Terminal words may have p r o p e r t i e s i n d i c a t i n g : 1) t h a t t h i s word is an acceptable value for some c l i n i c a l parameter(s) 2 ) t h a t this word always i m p l i c a t e s a certain c l i n i c a l parameter, system l i s t , o r table (e.g. t h e word \" i d e n ti t v l l always i m p l i c a t e s t h e parameter IDENTITY, which means t h e i d e n t i t y of an organism)3) t h a t this word might Xmplicate a c e r t a i n parameter, system l i s t , o r table (e.3. t h e word t l p o s i t i v e n might i m p l i c a t e t h e parameter NUMPOS, which means t h e number o f p o s i t i v e c u l t u r e s i n a series)4 ) t h a t t h i s word is part of a p h r a s e which can be thought of a s a sinqle word (examples of such phrases are \" t r a n s t r a c h e a l a s p i r a t i o n w , Ithow l o n g w , and llnot sterilev1.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF23": {
                "text": "Properties of' Terminal Words The first three p r o p e r t i e s are a c t u a l l y i n v e r s e p o i n t e r s which a r e generated a u t o m a t i c a l l y from p r o p e r t i e s of -the clinical parameters, S p e c i f i c a l l y , a word r e c e i v e s t h e \" a c c e p t a b l e valuew pointer t o a paramster ( p r o p e r t y ( 1 ) above) i l p it a p p e a r s i n t h e parameter's list o f acceptable values -a l i s t which i s used d u r i n g t h e c o n s u l t a t i o n t o check t h e u s e r ' s response t o 3 request for t h e parameter's value. Also, each c l i n i c a l oarameter, l i s t , and t a b l e has an a s s o c i a t e d list of key words t h a t are commonly used when t a l k i n g about this parameter, l i s t , or table. These words are d i v i d e d according t o hoM s u r e we can b e that a d o c t o r is referring t o this parameter list, or table when t h e articular word is used i n a q u e s t i o n . It is from t h i s l i s t t h a t t e r m i n a l words n i m p l i c a t i o n n pointers (~P a p e r t i e a 2 and 3 i n Table 1 ) are generated. During t h e f i r s t phase o f parsing, e a c h word i n t h e o r i g i n a l t e x t i s replaced by its t e r m i n a l word. For words n o t found i n the d i c t i o n a r y , t h e system uses Winograd's r o o t -e x t r a c t i o n a l e o r i t h m 181 t o see i f t h e word's l e x i c a l root is i n t h e d i c t i o n a r y ( e m s . , the root o f t f d e c i s i o n w is \" d e c i d e n ) . I f s o , t h e word is replaced by t h e t e r m i n a l word f o r i t s r o o t . Words st111 u n r e w g n i z e d aFter r o o t e x t r a c t i o n are l e f t unchanqed. The r e s u l t i n g l i s t o f t e r m i n a l and unrecognized words i s then passed t o a f u n c t i o n which r e c o g n i z e s phrases. Usinq p r o p e r t v 4 (see Table 1 ) o f the t e r m i n a l words i n t h i s l i s t , the function identifies a phrase and r e p l a c e s it w i t h a s i n g l e synonymous t e r m i n a l word (whose d i c t i o n a r y p r o p e r t i e s may be important i n determining t h e meaning of t h e q u e s t i o n ) . Classifying the Question The next s t e p is t o classify t h e q u e s t i o n s o that t h e proqram can t e l l which s p e c i a l i s t s h o u l d answer it. S i n c e a l l a u e s t i o n s about the c o n s u l t a t i o n must be about some s p e c i f i c c o n t e x t , t h e system r e q u i r e s t h a t the name of the context ( e . g . , ORGANISM-1) be s t a t e d explicitly. This ~i v e s a n easy way t o separate qeneral q u e s t i o n s about t h e knowledqe base from q u e s t i o n s about a p a r t i c u l a r c o n s u l t a t i o n . F u r t h e r c l a s s i f i c a t i o n is done through a p a t t e r n matchinp; approach siailar to t h a t used by Colby [ I 1. The l i s t o f words created by the first phase i s t e s t e d against a number of p a t t e r n s (about 50 a t present). Each p a t t e r n has a l i s t of a c t i o n s t o b e taken i f t h e p a t t e r n is matched. These a c t i o n s set f l a g s which i n d i c a t e what t y p e of q u e s t i o n was aaked. I n t h e case o f q u e s t i o n s about j u d q e n t a l knowledge (called rule-retrieval q u e s t i o n s ) , p a t t e r n matching also d i v i d e s t h e question i n t o t h e part r e f e r r i n g t o t h e r u l e ' s PREMISE and t h e part r e f e r r i n g to i t s ACTION. For example, i n \"How d o you decide t h a t a n organism is s t r e p t o c o c c u s ?~, there is no PREMISE p a r t , and t h e ACTION p a r t i s \"an organism is s t r e p t o c o c c u s~; i n \"Do you e v e r use t h es i t e of t h e c u l t u r e t o determine a n organism's i d e n t i t y ? \" , t h e PREMISE p a r tis \" t h e s i t e of t h e c u l t u r e w and t h e ACTION p a r t i s \"an organism's i d e n t i t y n .",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF24": {
                "text": "c l a s s i f i c a t i o n of a q u e s t i o n g u i d e s i t s f u r t h e r a n a l y s i s , Each q u e s t i o n t y p e has an a s s o c i a t e d template with blanks t o be f i l l e d i n from t h e q u e s t i o n , The d i f f e r e n t blanks and t h e t e c h n i q u e s f o r f i l l i n g them i n are l i s t e d i n Table 2. With t h e q u e s t i o n c o r r e c t l y c l a s s i f i e d , t h e g e n e r a l question-ans*erer can t e l l which s o e c i a l i s t s h o u l d answer it.F i l l i n 5 i n all blanks i n the template gives t h e s~e c i a l i s t a l l t h e information needed t o f i n d t h e answer, 1 ) < c n t x t > -The c o n t e x t must be mentioned by name.2 ) < r u l e > -E i t h e r a rule's name (RULE047) w i l l be mentioned, o r t h e word \" r u l e n w i l l a p p e a r , t o q e t h e r wiLh t h e rule 'a number ( 47).3 ) <value> -One o f t h e t e r m i n a l words i n t h e q u e s t i o nhas a d i a t i o n a r y p r o p e r t y i n d i c a t i n g t h a t i t i s a legal value for t h e parameter ( p r o p e r t y 1 , Table 1 -e , q . THRDAT is a legal value f o r t h e parameter SITE). 4 ) <parm> -A l l o f the words i n the l i s t are examined t o see i f t h e y i m p l i c a t e any c l i n i c a l p a r a m e t e r s , S t r o n g i m p l i c a t i a n s come from words w i t h p r o p e r t i e s showinrf t h a t t h e word is an a c c e p t a b l e v a l u e of the p a r a m e t e r , o r t h a t t h e word always i m p l i c a t e s t h a t parameter ( p r o p e r t i e s 1 and 2 , Table 1 ) . Weak i m p l i c a t i o n s come from words with n r o p e r t i e s showing that: t h e y might implicate t h e parameter ( p r o p e r t y 3, Table 1 ) . The system uses an e m p i r i c a l s c o r i n g mechanism f o r p i c k i n g o u t o n l y t h e most likely parameters. Associated w i t h c e r t a i n parameters are words o r p a t t e r n s which must a p p e a r i n t h e q u e s t i o n i n o r d e r f o r t h e parameter t o be implicated. T h i s scheme allows t h e system t o d i s t i n g u i s h related parameters which may be implicated by t h e same k e y words i n t h e first pass. For example, t h e word nPMNvf implicates parameters CSFPOLY (the percent of PMNs i n t h e CSF) and PMN ( t h e p e r c e n t o f PMNs i n t h e complete blood c o u n t ) . These are d i s t i n g u i s h e d by reauirinq t h a t t h e word \"CSFn be p r e s e n t i n a q u e s t i o n i n o r d e r f o r CSFPOLY t o be implic.at e d . 5 < l i s t > -System l i s t s are i n d i c a t e d i n a manner slmilar t o p a r a m e t e r s , except t h a t s c o r i n g i s n o t done. Lists, l i k e parameters, ,may have a s s o c i a t e d p a t t e r n s which must be p r e s e n t i n t h e q u e s t i o n . Furthermore, l i s t s h~v e properties t e l l i n g which other system lists are their s u b s e t s . If a q u e s t i o n i m p l i c a t e s both a l i s t and a subset o f t h a t l i s t , t h e more general ( l a r g e r ) l i s t i s d i s c a r d e d . n o g l y c o~i d e s ?~~ i m p l i c a t e s two lists: The l i s t o f a l l d r u g s and t h e l i s t o f drugs which are aminoglycosides. The system only c o n s i d e r s t h e more specific list o f aminoglycosides when answering t h e q u e s t i o n . 6 ) < t a b l e > -T a b l e s are i n d i c a t e d i n a manner similar t o l i s t s e x c e p t t h a t a n e n t r y i n t h e table must a l s o be resent i n t h e q u e s t i o n . F o r e x a m~l e , t h e word norganismw may indicate two tables: one c o n t a i n i n g a c l a s s i f i c a t i o n of organisms, and t h e other c o n t a i n i n g normal flora o f v a r i o u s p o r t a l s . The q u e s t i o n \"What organisms are c o n s i d e r e d t o b' e s u b t y p e s o f Pseudornona~?~' w i l l c o r r e c t l v i m p l i c a t e t h e former t a b l e , and \"What are the organisms likely t o be found i n t h e throat?ft w i l l i m p l i c a t e t h e l a t t e r , because FSEUDOMONAS is i n t h e first table and THROAT i s i n the second. Mechanisms f o r Analyzing a Question ** WHEN DO YOU DECIDE THAT AN ORGANISM IS A CONTAMINANT? [ I j Terminal words: WHEN DO YOU CONCLUDE THAT A ORGANISM IS l u e s : (CONTAMINANT (ANY) 4) (FORM (ANY) 1 ) /, A c t i o n ) (SAMEBUG ( A N Y ) 1 ) (COVERFOR ( A N Y ) 1 ) 14 3 F i n a l translation : Preaise: ANY A c t i o n : ( CONTAMINANT ANY ) 151 The r u l e s listed below conclude about: whether t h e organism is a contaminant",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF25": {
                "text": "It is d e f i n i t e t h a t the i d e n t i t y of the organism is one of: staphy7~coccus-coag-neg b a c i l l u ssubt ilis corynebacter ium-non-diphtheriaeThen: There is s t r o n q l y suqqestive evidence ( . 8 ) t h a t t h e orqaniqm is a contaminant Sample of MYCIN'S Analysis of a Question [ U s e r i n p u t follows the double asterisks. 1 [ 1 The quest ion is reduced to a l i s t of terminal w o r d s . parameters, the f i n a l translation t e l l s what rules can answer t h e question: there are no restrictions on the PREMISE, and t h e ACTION must contain the parameter CONTAMINANT ( w i t h any v a l u e ) . [ ] The answer c o n 8 f s t s of findinp; a l l rules which m e e t t h e s e restrictions, and p r i n t i n g t h o s e that the user wants to see.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF26": {
                "text": "each question type, there are a number of' possible answer t e m p l a t e s . For example, for questions of the fom ttHow do you know the value o f <parm> o f <cntxt>?\", two o f the answer templates are: I used <rule> t o conclude t h a t <parm> of < c n t x t > is <value>, This gave a cumulative c.f. of ( c e r t a i n t y factor), The last q u e s t i o n asked before t h e c o n c l u s i o n was made was (question number>. I n answer t o q u e s t i o n ( q u e s t i o n number) you s a i d t h a t <parm> of < c n t x t > i s <value> The s p e c i a l i s t f o r answering ,questions of a g i v e n tvpe w i l l need t o check t h e h i s t o r y tree or the system s knowledge base i n o r d e r t o determine which o f t h e answer templates i s a p p r o p r i a t e for a p a r t i c u l a r q u e s t i o n , Some blanks i n t h e answer t e m p l a t e will be filled i n by t h e same items t h a t f i l l e d blanks i n t h e q u e s t i o n tern~late. The remainder are f i l l e d by the specialist with information which w i l l answer t h e question. I n t h e above example, t h e s l o t s <parrn>, < c n t x t > , and p o s s i b l v <value> would be f i l l e d i n from the question, and the other slots would be f l l l e d from t h e h i s t o r v The s p e c i a l i s t t h a t provides answers t o such q u e s t i o n s must l o o k up t h e d e s i r e d information in the system's s t a t i c knowledge base. Generally, a n answer c o n s i s t s o f simply displaying t h i s i n f o r m a t i o n . As Figure 9",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF27": {
                "text": "s t r a t e s , t h e most processing that i s ever done is t o check the retrieved information against some item i n the question (for e a u a l i t y of set membership). ** I S BLOOD A STERILE SITE? Yes. Blood i s one of t h o s e s i t e s that are n o r m a l l y s t e r i l e . ** WHAT ARE THE NONSTERLLE SITES? Those sites t h a t are normally nonsterile are: CERVIX, CUTANEOUS-ULCER, LOCHIA, NOSE, SKIN, STOOL, THROAT, URETHRA, VAGINA, ** WHAT ORGANISMS ARE LIKELY TO BE FOUND I N THE THROAT? The 1 i k e l y ath hog ens a s s o c i a t e d with t h r o a t are : STREPT0C)OCCUS-PNEUMONIAE, STREPTOCOCCUS-GROUP-A, NFISSERIA-YGYINGITIDIS. ** I S BACTEROIDES AEROBIC? No, t h e a e r o b i c i t y of b a c t e r o i d e s i s a n a e r o b i c . * * WHAT DOSAGE OF STREPTOMYCIN DO YOU GENERALLY RECOMMEND? The s u g g e s t e d dosage and r o u t e o f s t r e p t o m y c i n i s 7,5 ma/kq q12h IM ( o r IV). ** ROW DO YOU TREAT MENINGOCOCCAL BACTEREMIA? For t r e a t m e n t of NEISSERIA-MENINGITIDIS i n PRIMARY-BACTEREMIA S e l e c t t h e r a p y from among t h e f o l l o w i n g d r u g s (ranked a c c o r d i n g l o relative e f f i c a c y , assuming uniform s e n s i t i v i t y ) : e : 60 rnq/kg I V t h e n 15 ms/kg q4h I V for 48 days t h e n 15 mq/kq q4h PO CHLORAMPHENICOL (a t i o n of therapy: 14 days Questiorrs a b o u t Factual Knowledge The s p e c i a l i s t t h z t answers q u e s t i o n s a b o u t j~dqUIenta1 knowledge i s s l i g h t l y !nore complicated. Answering t h e s e q u e s t i o n s ( F l q u r e 10) i n v o l v e sb u i l t -i n knowledge about t h e r u l e set, p l u s t h e a b i l i t y t o tlreadll t h e rules.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF28": {
                "text": "time t h e question has been a n a l y z e d , the m e c i a l i s t knows exactly which p a r a m e t e r s must appear i n the PREMISE, and which must apDear i n theACTION  of any rule which answers t h e q u e s t i o n . Values may be s p e c i f i e d f o r any of t h e parameters. To answer t h e question, t h e r u l e -r e t r i e v a l s p e c i a l i s t must f i r s t f i n d e v e r y rule whose PREMISE and ACTION satisfy t h e s e constraints. To do t h i s , i t needs t o know t h a t there are two special lists associated with each parameter: one contain in^ every rule that uses the parameter i n its PREMISE, and the other containinq every rule t h a t concludes about the parameter i n its ACTION. Using these l i s t s f o r t h e various parameters mentioned i n the quest ion, the s p e c i a l i s t can find those rules that might answer the question. I f no values were soecified, the job is done a n d t h e relevant rules can b e displayed without further anslvsis; otherwise, i t i s necessary t o read each of the rules i n t h e l i s t and t o eliminate those which do n o t mention the correct values f o r t h e ~arameter, The rule-retrieval specialist also makes use of a piece of MYCIN'S knowledge which was not discussed e a r l i e r . The system contains models of i t s own knowledge (called rule models) which are used primarily during acquisition of new medical knowledge from an expert [4]. These models, however, can be put to many uses -one i s t o explain aeneral patterns i n decision making. The rule models are abstract descriptions of a subsets o f rules and are generated automatically by readinq the rules. For examole, t h e model for IDENT-IS-PSEUDOMONAS tells what features are common t o the majority of rules which conclude t h a t t h e identity of an oraanism i s pseudomonas . I f a model exists describing t h e rules about which the question i s asking, t h e rule-retrieval soecialist incorporates t h i s model's information i n t o i t s answer (Figure 11). Thus the question-answering facility is able t o give some information about strategies for achieving some of the svstem's goals, a s well as t h e individual rules which use t h e strateqies. ** WHY DO YOU ASK WHETHER THE PATIENT HAS A FEVFR OF UNKNOWN ORIGIN? The s i t e of the culture is blood and 2J The number of cultures in the series i n c l u d i n g the c u l t u r e is qreater than or equal to 1 , and 3 ) The number of culturea i n t h i s series which were p o s i t i v e f o r the orqanism is less than or equal to 1, and 4) The p a t i e n t has a t r u e fever of unknown o r i g i n , and 5) \"ardiac-surgery is a r e l e v a n t item from the h i s t o r y of t h e p a t i e n t Then: There is suggestive evidence ( ' 6 ) t h a t t h e i n f e c t i o n is infective-endocardit is ** WHAT DOES GIJ MANIPULATION TELL YOU ABOUT THE IDENTITY OF AN ORGANISM? The rules l i s t e d below use: whether t h e p a t i e n t has had a s e n i t o -u r i n a r y manipulative procedure to conclude about: the identity o f t h e organism 156, 163, 190 Wfiich do you wish to see? The The morphology of t h e orqanism is rod, and4 ) The portal of entry of t h e organism is u r i n e , and 5) The p a t i e n t has n o t had a qenito-urinary manipulative procedure, and 6) C y s t i t i s is not a problem for which the p a t i e n t has been treatel? Then: There is suggestive evidence (.6) t h a t the i d e n t i t y of t h e orqanism 1s e. coLi",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF29": {
                "text": "r e 10. -Rule-Retrieval Questions ** HOW DO YOU DECIDE THAT AN ORGANISM MIGHT BE PSEUDOMONAS AERUGINOSAQ Rules which conclude that t h e i d e n t i t y of t h e organism i s pseudomonas-aeruginosa generally use one o r more of t h e f o l l o w i n g p i e a e s of i n f o r m a t i o n : t h e sfte of t h e c u l t u r e the Gram stain o f the organism t h e morphology of the oreanism Furthermore, t h e f o l l o w i n g r e l a t i o n s h i p s hold: The gram s t a i n of t h e orsanism, and the morphology of t h e organism tend t o appedr t o g e t h e r i n these r u l e s . RULE184, RULE1 16, RULE047, RULE085, RULE040 conclude that the identity of t h e organism i s pseudomonas-aeruqinosa. Which o f these do you wish t o see? The c a t e g o r y o f t h e organism i s n o t known, and The aram s t a i n of t h e o r s a n i s m i s qrarnneg, and 3 ) The morphology o f t h e organism i s r o d , and 4 ) The a e r o b i c i t y of t h e organism i s f a c u l Then: There is weakly s u g q e s t l v e evidence ( . I ) that t h e i d e n t i t y of the orqanism i s pseudomonas-aeruqinosa Question Which Uses Rule Models 2.5.2 Consultation-Specific Q u e s t i o n s One of t h e s i m p l e s t quest i o n s about a spec i f i c c o n s u l t a t i o n i n q u i r e s about the v a l u e o f a p a r t i c u l a r parameter (Figure 1 2 ) . The specialist which answers these q u e s t i o n s must know how to r e t r i e v e this i n f o r m a t i o n , I n TO WHAT CLASS DOES ORGANISM-1 BELONG? The category of ORGANISM-1 is e n t e r o b a c t e r i a c e a e ( '9) #* IS ORGANISM-1 A STREPTOCOCCUS? No. The i d e n t i t v of ORGANISM-1 i s e.coli 1.74) b a c t e r o i d e s (.7) klebaiella-pneumoniae (.4) ~seudomonas-aeruginosa (.22) e n t e r o b a c t e r (.21) proteus-non-mirabilis (.19) s e r r a t i a (.O3) salmonella (.02) Flgure 12. Q u e s t i o n s a b o u t a parameter's Value Answerinq o t h e r tllrpes of c o n s u l t a t i o n -s p e c i f i c questions i n v o l v e s the use o f knowledge about t h e control s t r b c t u r e , i n c l u d i n g : 1) how t h e system a c q u i r e s i n f o r m a t i o n why t h e system tries to f i n d out about parameters 3 ) how i n f o r m a t i o n i s used once i t is acquired 4) what causes a rule t o be t r i e d 5 ) what causes a rule t o f a i l , The s p e c i a l i s t f o r answering q u e s t i o n s l i k e !!How do vou know t h e value of <parm> of <cntxt>?w, knows that t h e value of a narameter can come from two sources: it can be deduced by r u l e s , o r the user can t e l l u s about it i n r e s p o n s e t o a question. The h i s t o r y tree (see S e c t i o n 2.2) w i l l show whlch ( p o s s i b l y b o t h ) o f these s o u r c e s provided t h e p a r t i c u l a r i n f o r m a t t o n mentioned i n t h e q u e s t i o n ( F i g u r e 13). ** HOW DO YOU KNOW THAT CULTURE-1 WAS FROM A STERILE SOURCE? I used rule189 t o conclude t h a t t h i s blood c u l t u r e was t a k e n from a s t e r i l e source. T h i s gave a cumulative c.f. of ( 1 . 0 ) . The last question asked before t h e conclusion was made was 7, ** DID YOU CONSIDER BACTEROIDES AS A POSSIBILITY FOR ORGANISM-I? Yes. I used rule095 t o conclude t h a t t h e i d e n t i t y of ORGANISM-1 i s b a c t e r o i d e s . This gave a curdulative c . f . of ( , 7 ) . The l a s t q u e s t i o n asked b e f o r e the conclusion was made was 20. ** HOW DO YOU KNOW THAT ORGANISM-1 IS E.COLI? The f o l l o w i n g r u l e s made conclusions about whether t h e i d e n t i t y of ORGANISM-1 i s e . c o l i cumulative certaimy l a s t q u e s t ion asked answer t o q u e s t i o n 9 you s a i d t h a t t h e i d e n t i t y of ORGANISM-1 i s e , c o l i ( , 3 ) Question Regarding How a Conclusion Was Made If t h e q u e s t i o n is phrased i n the negative, i t is neeessarv first t o f i n d all t h e ways t h e conclusion could have been made ( t h i s is a simple t a s k of rule-retrieval), then t o explain why it wasn't made in t h i s consuLtation (Figure 1 4 ) . The m e c i a l i s t f o r answerinq these questions must know what situations can prevent conclusions from beinq made. The second question i~",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF30": {
                "text": "illustrates how the answer+ t o one q u e s t i o n a i s h t cause another q u e s t i o n t o be asked.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF31": {
                "text": "Knowing t h i s with certainty r u l e s out all other values f o r the t h e morphology o f 0RQ.ANISM-1, i n c l u d i n q coccus.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF32": {
                "text": "Questions Regarding Why a Conclusion wasn't Made The specialist f o r answerinq q u e s t i o n s o f the fom \"How d i d vou use <parm> of <cntxt>?IV, needs t o know not o n l y how t o f i n d t h e s v e c i f i c rules which might u s e a parameter, but also how a ~a r a m e t e r can cause a r u l e t o f a i l 2nd how one parameter can prevent a n o t h e r from being used. The history tree can be checked t o see which of t h e relevant r u l e s used the oarameter, which f a i l e d because of t h e parameter, and which f a i l e d f o r some other reason, preventin@ t h e parameter from being used (Figure 15).** HOW D I D YOU USE THE AEROBICITY OF ORGANISM-17The a e r o b i c i t y of ORGANISM-1 was used i n the f o l l o w i n g rules:RULE003, RULE005, RULE084.The aembicity of ORGANISM-1 caused t h e f o l l o w i n g rules t o fail: RULEO35, RULEO51, RULE052, RULEO'i3, RULE111c Thc a e r o b i c i t y of ORGANISM-1 a l s o would have been used i n : RULE037, RULEOSO. RULE058, RULE086, RULE110, RULE184, RULE203, RULE204, RULE205. However, none o f t h e s e succeeded I n t h e c o n t e x t of ORGANISM-1. If you would l i k e an e x p l a n a t i o n For why any of t h e s e r u l e s failed, please e n t e r t h e i r numbers: ** 37 Rule037 was tried i n the context of ORGANISM-1, but it f a i l e d due t o c l a u s e 1 [ \" t h e c a t e g o r v of t h e organism i s n o t known\"] . ** D I D YOU CONSIDER THE FACT THAT PATIENT-1 I S NOT A COMPROMISED HOST? The f a c t that Fred Jones i s not a compromised h o s t caused t h e following rules t o fail: RULE085, RULE106. The f a c t t h a t ~r e a Jones is n o t a compromised h o s t a l s o would have been, used i n : RULE109. Howeveri none of these succeeded i n the context of PATIENT-1.. I f you would l i k e an e x p l a n a t i o n f o r why any of these r u l e s f a i l e d , * please enter t h e i r numbeps: ** NONE: Queathns Regarding How I n f o r m a t i o n Was Used For questions of t h e form \"Whv d l d n ' t you f i n d o u t about <parin> o f <cntxt>?.\", q e n e r a l knowledge of MYCIN'S c o n t r o l s t r u c t u r e t e l l s t h e c o n d i t i o n s under which it would have heen n e c e s s a r y t o f i n d out some piece o f i n f o r m a t i o n . The record o f t h e consultation can b e used t o d e t e r m i n e why t h e s e c o h d i t i o n s n e v e r arose f o r t h e p a r t i c u l a r parameter i n q u e s t i o n (Figure 16)-.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF33": {
                "text": "16 also illustrates t h a t MYCIN'S GQA allows a user t o g e t as much information a s is desired. The f i r s t answer given was not r e a l l y complete In i t s e l f , b u t it l e d t h e user t o ask a n o t h e r q u e s t i o n t o y e t more i n f o r m a t i o n . Then another questlgn was asked t o d e t e r m i n e why clause 1 or R U L E I~~ was false. The answers t o t h e first two questions both mentioned rules which could be printed i f t h e user wanted t o examine them. The special command PR ( P r i n t ~u l e ) ig for t h e user's convenience. It r e q u i r e s no natural-languaqe processing and thus can be understood and answered immediately. (What is rule or * ' P r i n t rule 109 .\" a l s o would have been underst oad . ) ** DID X8U FIND OUT A6OUT THE CBC ASSOCIATED WITH CULTURE-I? During t b e preceding c o n s u l t a t i o n , I n e v e r had r e a s o n t o try t o f i n d out whether t h e r e s u l t s of t h e CBC from t h e time p e r i o d o f t h i s blood c u l t u r e are a v a i l a b l e . It would have been n e c e s s a r y t o know t h i s i n f o r m a t i o n in o r d e r t o use any o f t h e following: RULE109. However, none o f these succeeded i n t h e c o n t e x t o f t h i s blood c u l t u r e . I f you would like a n e x p l a n a t i o n f o r why any o f t h e s e r u l e s f a i l e d , please e n t e r t h e i r numbers: ** 109 Rule109 would be relevant t o t h i s blood c u l t u r e when a p~l i e d t o : ORGANISM-1. I n e v e r had reason t o trv t o apply r u l e 1 0 9 t o ORGANISM-1. I would have needed t o u s e r u l e 1 0 9 i n o r d e r t o f i n d o u t whether ORGANISM-1 is a c o n t a b i n a n t . However, d u r i n g the p r e c e d i n g c o n s u l t a t i o n , 1 n e v e r had r e a s o n t o trv t o f i n d out whether ORGANISM-1 is a contaminant. ** WHY DIDN'T YOU NEED TO KNOW WHETHER ORGANISM-1 IS A COWTAMINANT? During t h e preceding ~o n a u l t a t i o n , 1 n e v e r had r e a s o n t o try t o f i n d o u t whether ORGANISM-1 i s a contaminant. It would have been necessarv t o know this i n f o r m a t i o n i n o r d e r t o use any of the followinq: RULEOO6, RULE007, RULE906, RULEt08, RULE109, RULl3159. However, none o f these succeed'ed i n t h e c o n t e x t o f ORGANISM-1. I f you would like an e x p l a n a t i o n f o r why any of these r u l e s failed, please e n t e r their numbers: ** 159 Rule159 was t r i e d i n the c o n t e x t of ORGANISM-1, b u t i t f a i l e d due t o clause 1 [\"it is suspected t h a t t h e i d e n t i t y of t h e organism i s corynebacterium-non-diphtheriaew 1 ** I S ORGANISM-1 CORYNEBACTERI UM-NOW-DIPHTRERIAE? No. The i d e d t i t y of ORGANISM-1 i s e , c o l i (.74) b a c t e r o i d e s (.7) klebsiella-pneumoniae (.4) pseudomonas-aeruqinosa ( 2 2 ) e n t e r o b a c t e r (.21) proteus-non-mirabilis (.I91 serratia (.03) salmonella f.02) The c u J t u r e was taken from a s t e r i l e s o u r c e , and Xt is not known whether tb e orqanism i s a contaminant, and 3 ) The patient is n o t 9 compromised h o s t , and 4) The p a t i e n t was f e b r i l a due t o t h e i n f e c t i o n , and 5 ) The r e s u l t s o f t h e cbc from t h e time ~e r i o d o f the culture are available, and 6 ) A -The white count from the CBC i s q r e a t e r t h a n 10.5, or B -The percent o f pmn's i n the cbc a t t h e time t h e culture was obtained is qreater t h a n 78, o r C -The p e r c e n t o f wbc's which were bands i n t h e CBC a t the time the, c u l t u r e was o b t a i n e d is q r e a t e r than 10 Then: There i s strongly suggestive evidence (.a) that the orqanism is n o t a contaminant F i g u r e 16. Q u e s t i o n s Regarding Why a Parameter wasn't Traced In questicns asking about the a p p l i c a t i o n of a r u l e t o a c o n t e x t there are three p o s s *~b i l i t i e s : the r u l e told u s somethinq about t h e c o n t e x t ; t h e rule f a i l e d when applier! t o that context; or t h e rule was never tried i n t h a t c o n t e x t , The h i s t o r v tree tells which of these i a t h e case, Furthermore, if a rule succeeded, there is a record of all the conclusions i t made, and i f it I\"3ileb, t h e reason f o r f a i l u r e i s recorded. 9s Fiqure 17 i l l u g t r a t e s , rules are o n l y a p p l i c a b l e i n c e r t a i n c o n t e x t s . I f t h e r u l e could not a p p l y t o t h e context mentioned i n t h e q u e s t i o n , b u t m i~h t be relevant when a p p l i e d t o soae o t h e r context [which is an ancestor o r descendent i n t h e c o n t e x t tree], then the q u e s t i o n i s answered a s i f it had",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "FIGREF34": {
                "text": "make such a d e c i s i o n , d e s p i t e t h e ~h r a s i n q o f the question. Similarlv,when asked how t h e system used ( o r concluded about) some mrameter t h a t was never traced, the system e x p l a i n s why it d i d n 't need t o f i n d out about t h e parameter, reflardless of how t h e question was bhrased.** DID RULE 116 TELL YOU ANYTHING ABOUT INFECTION-I?~ulell6 would be relevant to the primary-bacteremia when applied t : ORGANISM-1Clause 5 of rulel 16 [\"the infection is m e n i n q i t i s n ] was already known to be false for ORGANISM-I, so the rule was never t r i e d .** HOW WAS RULE. 178 HELPFUL WHEN YOU WERE CONSIDERING ORGANISM-13~u l e 1 7 8 would be relevant to ORGARISM-1 when applied to:CULTURE-I I never had reason t o try t o a p p l y rule178 t o t h i s bloodc u l t u r e . X would have needed to use r u l e 1 7 8 i n order t o find out whether t h i s blood culture was t a k e n from a sterile s o u r c e . However, I was already c e r t a i n t h a t t h i s blood culture was t&ken from a qterile source, ** WHAT DID RULE 295 TELL YOU ABOUT ORGANISM-I? I used rule295 t o conclude that the identity of ORGANISM-1 $3 hemophilus-influenzae. This gave a cumulative c.f. of (25). The l a s t q u e s t i o n asked before the conclusion w a s made was 36. +* WHY DIDN'T YOU USE.RULE112 TO FIND OUT ABOUT ORGANISM-I? RULE1 12 was not executed because it would have caused circular reasoning when a p p l i e d to ORGANISM-?. Would you like ta see t h e chain of rules and parameters which makes up t h i s circle? ** YES 1 wanted to know about the identity of ORGANISM-1 -because I t r y to find out the i d e n t i t v of the organism for a l l current organisms of t h e p a t i e n t . To find out about the i d e n t l t y of ORGANISM-1, I tried to use ruleO21. Before I could use rule021, I I trred t o use rule005. Before I could use rule005, I needed to know about the aerobicity of ORGANISM-1 . To firid out about the aerobicity of ORGANISM-1, I tried to use r u l e 0 3 1 . Before I could use rule031, I needed to know about the category of ORGANISM-1. T o f i n d out about the category of ORGANISM-1, I tried to use rulel 12. Before I could use rulel 12, I needed to know about the i d e n t i t y of ORGANISM-1 . But t h i s is the unknown parameter I sought o r i g i n a l l y . Figme 17. Question Regarding the Application of a Rule 3 Conclusions C o n s u l t a t i o n system8 which g i v e e x p e r t a d v i c e i n some domain form one class of a r t i f i c i a l i n t e l l i g e n c e programs which can provide useful solutions t o real-world ppoblems. The u t i l i t y of such a system, however, depends on its a c c e p t a b i l i t y t o human users. One feature which can Increase a system's a c c e p t a b i l i t y is a mechanism whereby the system can explain or j u s t i f y i t s a d v i c e . The development bf an explanation mechanism f o r a consultation system i s v e r y much related t o t h e problems o f r e p r e s e n t i n g knowledge and of making use of d i f f e r e n t s o u r c e s of knowledge, Since t h e production system formalism p r o v i d e s a unified way t o r e p r e s e n t modular pieces o f knowledae, the task of d e s i g n i n g an e x p l a n a t i o n c a p a b i l i t v is s i m p l i f i e d f o r production-based consultation systems, The example of MYCIN shows how t h i s can be done and i l l u s t r a t e s further t h a t a svsterll desiqned for a single domain with a small, t e c h n i c a l vocabulary can give comprehensive answers t o a wide range of q u e g t i o n s without s o p h i s t i c a t e d natural-language processing.",
                "num": null,
                "uris": null,
                "type_str": "figure"
            },
            "TABREF0": {
                "html": null,
                "text": "ContentsPerformance Character lstics of an E x~l a n a t i o nC a~a b i 1 i . t~ . . . . . . . . . . I . . . 7 Knowledge Requirements of an Explanation C a~a b i l i t v . . -9 . . . . . . . . . -",
                "content": "<table><tr><td>S e c t i o n</td><td/><td/><td/><td>Pane</td></tr><tr><td colspan=\"2\">Subsec t l o n</td><td/><td/></tr><tr><td colspan=\"2\">General Discussion</td><td>. . . . . .</td><td>. . \u2022</td><td>. 4</td></tr><tr><td/><td/><td/><td/><td>. 4</td></tr><tr><td>1.2</td><td/><td/><td/></tr><tr><td>1.3</td><td/><td/><td/></tr><tr><td>1 . 4</td><td/><td/><td/></tr><tr><td/><td/><td/><td/><td>. 16</td></tr><tr><td>2.1</td><td>Overv lew</td><td/><td/></tr><tr><td>2.2</td><td/><td/><td/></tr></table>",
                "type_str": "table",
                "num": null
            },
            "TABREF1": {
                "html": null,
                "text": "",
                "content": "<table><tr><td>O R I G I N ?</td></tr><tr><td>ORGAN ISM'S IDENTSTY?</td></tr></table>",
                "type_str": "table",
                "num": null
            },
            "TABREF2": {
                "html": null,
                "text": "",
                "content": "<table/>",
                "type_str": "table",
                "num": null
            }
        }
    }
}