| { |
| "paper_id": "Y00-1023", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T13:45:15.600620Z" |
| }, |
| "title": "DEVELOPING DATABASE SEMANTICS AS A COMPUTATIONAL MODEL*", |
| "authors": [ |
| { |
| "first": "Kiyong", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Korea University", |
| "location": { |
| "postCode": "136-701", |
| "settlement": "Seoul", |
| "country": "KOREA" |
| } |
| }, |
| "email": "klee@mail.korea.ac.kr" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "Both Hausser [1] and Lee [2][3] proposed Database Semantics as a computational model for natural language semantics that makes use of a database management system, DBMS. As an extension of these efforts, this paper aims at dealing with ways of representing linguistic descriptions in table forms because all the data in a relational model of DBMS is conceived of being stored in table forms. It is claimed here that, if an algorithm can be developed for converting linguistic representations like trees, logical formulas, and attribute-value matrices into table forms, many available tools for natural language processing can be efficiently utilized as part of interface or application programs for a relational database management system, RDBMS.", |
| "pdf_parse": { |
| "paper_id": "Y00-1023", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "Both Hausser [1] and Lee [2][3] proposed Database Semantics as a computational model for natural language semantics that makes use of a database management system, DBMS. As an extension of these efforts, this paper aims at dealing with ways of representing linguistic descriptions in table forms because all the data in a relational model of DBMS is conceived of being stored in table forms. It is claimed here that, if an algorithm can be developed for converting linguistic representations like trees, logical formulas, and attribute-value matrices into table forms, many available tools for natural language processing can be efficiently utilized as part of interface or application programs for a relational database management system, RDBMS.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "For business transactions or academic administration, a commercial database management system (DBMS) like DB2, Oracle or Informix is widely used. Database Semantics is an attempt to adopt such a system for doing semantics for ordinary language. Since an interpretation model or background is necessary for processing linguistic information, Database Semantics can use as its model a database in a DBMS that provides both lexical meaning and world knowledge information. Furthermore, a DBMS constantly updates its database with new data, as fragments of a natural language like Korean or English are processed through a linguistic processing system LIPS. Hence, Database Semantics can consistently process even a larger fragment of discourse in natural language.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "INTRODUCTION", |
| "sec_num": "1." |
| }, |
| { |
| "text": "(1) A Model of Database Semantics But, before developing Database Semantics as an application program into an RDBMS, the issue of representation must be resolved at least in the logical or conceptual level. Unlike ordinary semantics, Database Semantics as a computational model ultimately aims at a computational implementation. Hence, it should be able to build a successful interface between the two different levels of linguistic description and computational processing. As a result, representation comes to play an important role of bridging the and computational processing. As a result, representation comes to play an important role of bridging the gap between human and computer interactions. Linguistic descriptions are often represented in trees, logical forms, or attribute-value matrix (AVM) structures.' On the other hand, a computer system like RDBMS, which only knows a relational language like SQL, can conceptually recognize data stored in a table form only. This paper will thus focus on ways of representing linguistic information in a table form so that it can be recognized by an RDBMS.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "INTRODUCTION", |
| "sec_num": "1." |
| }, |
| { |
| "text": "The main task of this paper is then to discuss ways of converting linguistic representations like trees, logical forms, and matrices into table forms. If an algorithm is developed for these conversions, then many of the exiting parsers or interpreters can be directly incorporated into an RDBMS. Head-driven Phrase Structure, Lexical-Functional, and Left-Associative grammars, for instance, produce the results of their analysis represented in a sequence of Trie or AVM structures.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "INTRODUCTION", |
| "sec_num": "1." |
| }, |
| { |
| "text": "At the present, however, the presentation of such an algorithm is beyond the scope of this paper, for the version of Database Semantics proposed in the framework of RDBMS is still in the nascent state of being designed as a computational model. It is thus only hoped that fragmentary but concrete illustrations will be given to show how linguistic analyses can be represented in table forms.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "INTRODUCTION", |
| "sec_num": "1." |
| }, |
| { |
| "text": "Hausser [1] was the first to develop Database Semantics. By adopting a network model for DBMSs, he proposed to construct a Word Bank, a lexical database consisting of word types and their tokens. By navigating this Word Bank, propositional content can then be processed linearly or left-associatively by a step-by-step manner. In a similar vein, Lee [2] also showed how a network model could be used to represent various structural relations in natural language analysis, while assuming that it could be a minimally sufficient model for processing natural language.", |
| "cite_spans": [ |
| { |
| "start": 8, |
| "end": 11, |
| "text": "[1]", |
| "ref_id": "BIBREF0" |
| }, |
| { |
| "start": 350, |
| "end": 353, |
| "text": "[2]", |
| "ref_id": "BIBREF1" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY A RELATIONAL MODEL?", |
| "sec_num": "2." |
| }, |
| { |
| "text": "Lee [3] , however, argued for some advantages of adopting a relational model possibly with object extensions. First, the relational model has given a conceptual basis for implementing most of the currently running commercial DBMSs like Oracle8i or Informix7.0.", |
| "cite_spans": [ |
| { |
| "start": 4, |
| "end": 7, |
| "text": "[3]", |
| "ref_id": "BIBREF2" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY A RELATIONAL MODEL?", |
| "sec_num": "2." |
| }, |
| { |
| "text": "Secondly, the relational model uses standardized SQL, a structured query language, for describing ways of building database structures and managing them, thus making it possible to apply it to the construction of a query system for natural language semantics.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY A RELATIONAL MODEL?", |
| "sec_num": "2." |
| }, |
| { |
| "text": "Finally, its basic representational scheme at the logical or conceptual level is of a table form consisting of attributes and their values like an attribute-value matrix for linguistic representation. In the relational model, a database is a set of tables representing various relations among objects in a domain. This paper thus proposes to adopt a relational model RDBMS for developing Database Semantics.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY A RELATIONAL MODEL?", |
| "sec_num": "2." |
| }, |
| { |
| "text": "There are at least two reasons for adopting a DBMS as a basis for doing semantics. First, natural language generates a large amount of linguistic information with great complexity. Even a tiny fragment of text contains a lot of data. Especially when it contains indexical or context-dependent expressions, even a short sentence consisting of three or four words can theoretically create almost an infinite number of interpretations. Consider the following sentence:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "(2) I am happy today.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "The original version of this paper discussed the tripartite representation of quantified sentences in AVM structures and converting these matrices into table forms. But, due to the editorial constraint on the limitation of space, the discussion of AVM structures has to be postponed for another occasion. This string of words in English constitutes a well-formed sentence of English, meaning that, whoever the speaker is, she or he is happy on the day of her or his making the statement. Hence, it can have an indefinitely large number of interpretations depending on who the speaker is and also when or what day it is spoken.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "Let's consider another example. Suppose two lovers say to each other:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "(3) I love you.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "This sentence consists of only three simple words. But, depending on the context of its use, it can produce infinitely many different interpretations. Here, neither I nor you refers to a fixed person. Even in a situation where two persons, say Mia and Bill, say to each other I love you, each of the pronouns has two different referents. The pronoun I once refers to Mia, while the pronoun you refers to Bill. And then the pronoun I refers to Bill, whereas the pronoun you refers to Mia. Hence, a model in which their referents are fixed cannot interpret sentences like (3).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "By allowing a DBMS to keep updating its database, Database Semantics as a model-theoretic semantics should be able to use the constantly updated set of data supplied from the database to interpret contextually dependent statements or utterances. By appropriately partitioning the database into smaller parts through DBMS, Database Semantics can get the effect of subdividing its interpretation model into smaller sub-models so that the indexical expressions like I or you or even anaphoric pronouns like she or he may have varying referents within a larger model. Event statements have the same problem of varying reference. Indexical expressions like today or yesterday have different referents as discourse situations vary. Then there are events that consist of smaller events, some of which may overlap each other. While statements are being made continuously, their interpretation model should be able to keep track of their sequential as well as overlapping relations,. This task, Database Semantics is expected to perform efficiently by making a DBMS to control and manage its database.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "While using a database as its interpretation model, Database Semantics installs in it various lexical and grammatical modules for processing natural language. The lexicon or word bank should be part of the database. This lexicon may be of various types. It can, for instance, be a multi-lingual dictionary or may contain a virtual dictionary temporarily built to deal with a given fragment of discourse.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "The entire grammar consisting of both syntactic and semantic rules should also be installed in the same database. There may be sets of constraints or principles governing the generation of well-formed sentences. DBMS then manages and upgrades all these modules with new lexical items as well as newly required grammatical rules and constraints, as a larger linguistic corpus is introduced into the system. By allocating all these managing functions to DBMS, Database Semantics can make its module LIPS for linguistic information processing simply function as an application interface between its users and DBMS.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "WHY DATABASE SEMANTICS?", |
| "sec_num": "3." |
| }, |
| { |
| "text": "A relational DBMS is normally identified with tables.' A table is a very simple and rigid form of representation, a triplet <Name, Attributes, Values> consisting of a name, a set of attributes or fields and their corresponding values. Nevertheless, it can carry a lot of information with varying types. The following is an example containing some personal information on professors:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "2 More detailed discussions of the notion of table and its properties in an RDBMS can be found any ordinary books on database management systems like [4] , [5] , [6] , and [7] . ", |
| "cite_spans": [ |
| { |
| "start": 150, |
| "end": 153, |
| "text": "[4]", |
| "ref_id": "BIBREF3" |
| }, |
| { |
| "start": 156, |
| "end": 159, |
| "text": "[5]", |
| "ref_id": "BIBREF4" |
| }, |
| { |
| "start": 162, |
| "end": 165, |
| "text": "[6]", |
| "ref_id": null |
| }, |
| { |
| "start": 172, |
| "end": 175, |
| "text": "[7]", |
| "ref_id": "BIBREF6" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "This tree is understood as representing how sentence (8) is formed: it consists of a noun Mia, a verb loves, and another noun Kim.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "The same structural information can be represented in the following table form where the terminal nodes Mia, loves, and Kim are marked with a dummy symbol t:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "(9) Here, a unique code number is assigned to each node from which one can obtain information on the dominance and precedence relations among the nodes. The NODE_CODE 0, for instance, stands for the root node labeled S, and the NODE CODEs 01 and 02 are its two daughters, NP and VP respectively. It is again understood here that 0 dominates 01 and 02 and that 01 precedes 02. This is so because the number of digits in each NODE_CODE is understood as representing its level in the tree and also the sequence of numbers except for the one in the last digit in each NODE_CODE as representing the code of its mother.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "In Phrase Structural Grammar, the terminal nodes Mia, loves, and Kim in a tree like (8) are first marked with a dummy symbol like t, as in (9). It is then replaced by an appropriate lexical item selected from a lexicon by the process of lexical insertion. This fact can also be easily captured in a sequence of tables linked to each other. For this, we first construct a lexicon table like the following:", |
| "cite_spans": [ |
| { |
| "start": 84, |
| "end": 87, |
| "text": "(8)", |
| "ref_id": "BIBREF7" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "LEX_CODE CATEGORY WORD FORM n_l N Kim n2 N Mia v_l V hates v2 V loves", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Secondly , the TREE table can be linked to the LEXICON table through the following bridging table: NODE_CODE LEX_CODE 011 n_2 021 v2 0221 n_l This is a very rigid table, for it only allows the selection of a single word for each pre-terminal node. But it makes it possible to convey the exact amount of information as represented by tree (8) .", |
| "cite_spans": [ |
| { |
| "start": 346, |
| "end": 349, |
| "text": "(8)", |
| "ref_id": "BIBREF7" |
| } |
| ], |
| "ref_spans": [ |
| { |
| "start": 9, |
| "end": 145, |
| "text": ", the TREE table can be linked to the LEXICON table through the following bridging table: NODE_CODE LEX_CODE 011 n_2 021 v2 0221", |
| "ref_id": null |
| } |
| ], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "For generation, we need a more flexible way of building tables and linking them to each other. From the pre-terminal tree table (9), it should be possible to construct a table frame like the following:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "(12) It has been shown here how phrase structure information can be represented in a table form and also how the processes of language analysis and generation can be captured by a sequence of tables with appropriately constructed bridging tables. It should, however, be noted that the use of tables for representing constituent structures should not be understood as claiming the adequacy of Phrase Structure Grammar for linguistic description. It rather shows the possibility of accommodating a grammar system that generates trees into an RDBMS as a genuine part of the LIPS of Database Semantics.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "TABLES", |
| "sec_num": "4." |
| }, |
| { |
| "text": "In formal semantics for natural language, sentences are first disambiguated through syntactic analysis and then translated into a logical language like Intensional Logic for their model-theoretic interpretations. In this section, I will show how logical formulas can be converted to tables, which represent the propositional content of sentences in natural language.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "FROM LOGICAL FORMULAS TO TABLES", |
| "sec_num": "6." |
| }, |
| { |
| "text": "Simple sentences are normally translated into atomic formulas in a Predicate Logic-like language, which consists of a relation and a set of its arguments. Suppose we have a simple sentence:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(15) Mia loves Kim.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "This sentence can be translated into an atomic formula like the following:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "EQUATION", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [ |
| { |
| "start": 0, |
| "end": 8, |
| "text": "EQUATION", |
| "ref_id": "EQREF", |
| "raw_str": "(16) love'(mia', kim')", |
| "eq_num": "(14)" |
| } |
| ], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "This formula consists of a relation love' and two arguments mia' and kim'.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "This formula can easily be turned into a Here, the embedded sentence Kim loves her plays its role as the Object of the verb knows.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "This sentence may have a logical representation like the following:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(21) knows'(mia', loves'(kim', mia))", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "The predicate knows' takes two arguments: one is an individual constant mia' and the other a proposition loves'(kim', mia').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "By adopting Lee's equation-solving approach [8] , this formula may be decomposed into two atomic formulas as in:", |
| "cite_spans": [ |
| { |
| "start": 44, |
| "end": 47, |
| "text": "[8]", |
| "ref_id": "BIBREF7" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(22) knows'(mia', p) p = (loves'(kim', mia'))", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "EQUATION", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [ |
| { |
| "start": 0, |
| "end": 8, |
| "text": "EQUATION", |
| "ref_id": "EQREF", |
| "raw_str": "(17)", |
| "eq_num": "(18)" |
| } |
| ], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "The letter p is used here as a variable of the propositional type. It is equated in (22) to the proposition loves'(kim', mia').", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "A more accurate way to represent the use of the pronoun her in sentence (20) is to represent it as a variable because it may not necessarily refer to Mia. Only if some background is provided appropriately, it can be equated to mia' and then refer to Mia. Otherwise, it is left unbound as a free variable.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "By accepting such a treatment of pronouns as variables, translation (21) can be reformulated as in (23):", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(23) knows'(mia', p) p = (loves'(kim', x') x = mia'", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "This translation consists of a set of an atomic formula and its related equations. One equation refers to the content of the relation knows' and the other binds the variable x to mia'.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "These formulas can also be converted to a table like the following: The logical form of this sentence may look like the following:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(24)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(28) PAST(promise'(mia', kim', help'(mia',kim')))", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "Through Lee's equation-solving approach [8] , the above representation can be decomposed into the following set of equations with a propositonal formula.", |
| "cite_spans": [ |
| { |
| "start": 40, |
| "end": 43, |
| "text": "[8]", |
| "ref_id": "BIBREF7" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "(29) PAST(promise'(mia', kim', p)) p = help'(x, y)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "x= mia', y = kim'", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "The equation x=mia' is supported by the so-called EQI-NP constraint or some control mechanism associated with the verb promise in English, while the equation y=kim' is introduced by some background information that governs the use of pronouns.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "In order to represent these pieces of information in a Table ( 30) is linked to tables (31) and (32) through the keys supplied by PROP_CODE, BG_CODE, and CL CODE in these tables.", |
| "cite_spans": [], |
| "ref_spans": [ |
| { |
| "start": 55, |
| "end": 62, |
| "text": "Table (", |
| "ref_id": "TABREF4" |
| } |
| ], |
| "eq_spans": [], |
| "section": "SIMPLE SENTENCES", |
| "sec_num": "6.1" |
| }, |
| { |
| "text": "In natural language, each verb must carry information on tense. In a coordinate or complex sentence, some verbs may lack any overt tense marking, as with the verb marry in the following sentence:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "(33) Bill promised Mia to marry her.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "Here the event of marrying is understood to take place not before, but after the promise was made. This interpretation is possible because any promised event must occur later than the time of making the promise itself. Consider the following pair of sentences:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "(34) a. Bill promised that he married her. b. Bill promised that he would marry her.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "Both sentences are syntactically acceptable. Unlike (34b), however, sentence (34a) cannot be properly interpreted semantically.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "(30 (3 1(32 By borrowing the representation scheme of neo-Davidsonian Event Semantics, the content of sentence (33) and its background may be captured as in:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "The value temp_2 of T_CODE shows that el precedes e2.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "The third row prop_3 in the PROP (35) promise'(e\" bill', mia', p) p = marry'(e2 , x, y) PAST({ei}) precedes'(ei,e2) x = bill', y = mia'", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "Here are two events, e l and e2 : one is an event of Bill's making a promise to Mia, and another, an event of marrying. The event e l is a past event. Hence, it is represented as part of the interval PAST. The predicate promise' relates two events or actions by arranging the temporal sequence of their occurrences. The formula precedes'(e\"e2) states that the event of promising occurs before the event of the promised event or action.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "Two equations for checking coreference are again introduced here. Since the verb promise is a subjectcontrol verb, the invisible subject of the infinitival clause is understood to be the same as the main subject.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "The equation x = bill' expresses this fact. The second equation y = mia' again depends on presumable background information, which thus makes it possible to interpret the pronoun her as referring to Mia.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "The semantic information represented by formulas in (35) will be stored separately in four different tables for PROP, TEMP_REL, CONTROL, and BACKGROUND information. The PROP ", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "On the basis of this BACKGROUND table, the variable y marked with the value bg_9 of BG_CODE is equated to mia'.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "The third row also contains information on CL CODE. This again tells the invisible subject of the infinitival complement of the verb promise is controlled to be coreferential with the main subject Bill.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "In this section, it has been shown how the propositional content of simple and complex sentences and their related background or control information can be represented in table forms and also how these tables can be linked to one another to share some information. As should be noted again, these representations are only conceptual constructs. Hence, further refinements may become necessary at the physical level of implementation.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "EVENT-RELATED STATEMENTS", |
| "sec_num": "6.3" |
| }, |
| { |
| "text": "A version of Database Semantics proposed here is based on an RDBMS. It is first characterized by the representation of data in table forms and secondly by a relational query language SQL. In this paper, nothing is discussed on the use of SQL for developing a query system of NL understanding or generation. Instead, efforts were made to discuss the use of tables for linguistic description.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "CONCLUDING REMARKS", |
| "sec_num": "7." |
| }, |
| { |
| "text": "The use of tables in linguistics is not totally unknown. Ordinary grammar books and dictionaries contain tables of verbal inflections and nominal declensions. Allomorphic variations and irregular and exceptional word forms are often listed in table forms. In computational work, too, we find a technique like Chart Parsing that uses tables. The use of tables in Database Semantics may be understood as a formal way of extending such traditional use to NL semantics.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "CONCLUDING REMARKS", |
| "sec_num": "7." |
| }, |
| { |
| "text": "The use of tables in Database Semantics is simply a consequence of adopting a relational DBMS. In an RDBMS, data is conceptually viewed as being stored in table forms. Although they are pure conceptual constructs, tables are meta-theoretic objects to which mathematically explicit operations can apply. Hence, to build a theoretically adequate version of Computational Semantics, a design mechanism based on such constructs may be of great use.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "CONCLUDING REMARKS", |
| "sec_num": "7." |
| }, |
| { |
| "text": "The use of tables in currently running commercial RDBMSs is very much restricted. It can, however, gain a more expressive power for linguistic description through object extensions and further implementation. Different types of data or records are exchanged or augmented through primary or foreign keys that are specified in tables and also as part of linking rules. If such a manner of linking from row to row can be extended to a new manner of linking from table to table, then it must be a greatly desirable improvement. For linguistic analysis often requires globally linked blocks of information.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "CONCLUDING REMARKS", |
| "sec_num": "7." |
| } |
| ], |
| "back_matter": [], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "Foundations of Computational Linguistics: Man-Machine Communication in Natural Language", |
| "authors": [ |
| { |
| "first": "Roland", |
| "middle": [], |
| "last": "Hausser", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Roland Hausser. Foundations of Computational Linguistics: Man-Machine Communication in Natural Language, Berlin: Springer-Verlag, 1999.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Formal Semantics and Description of Korean", |
| "authors": [ |
| { |
| "first": "Kiyong", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "" |
| }, |
| { |
| "first": ";", |
| "middle": [], |
| "last": "Beom-Mo Kang", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "498--551", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kiyong Lee. \"Computational Linguistics\" [written in Korean], in Beom-mo Kang et al., Formal Semantics and Description of Korean [written in Korean], pp. 498-551, Seoul: Hanshin, 1999.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Language and Representation: A Basis of Database Semantics", |
| "authors": [ |
| { |
| "first": "Kiyong", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "Proceedings of the Eleventh Conference on Hangul and the Korean Language Information Processing", |
| "volume": "", |
| "issue": "", |
| "pages": "297--303", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kiyong Lee. \"Language and Representation: A Basis of Database Semantics\" [written in Korean], Proceedings of the Eleventh Conference on Hangul and the Korean Language Information Processing, pp. 297-303, 1999.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Foundation for Object/Relational Databases: The Third Manifesto", |
| "authors": [ |
| { |
| "first": "C", |
| "middle": [ |
| "J" |
| ], |
| "last": "Date", |
| "suffix": "" |
| }, |
| { |
| "first": "Hugh", |
| "middle": [], |
| "last": "Darwen", |
| "suffix": "" |
| } |
| ], |
| "year": 1998, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "C.J. Date and Hugh Darwen. Foundation for Object/Relational Databases: The Third Manifesto, Reading, MA: Addison-Wesley, 1998.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Fundamentals of Database Systems", |
| "authors": [ |
| { |
| "first": "Ramez", |
| "middle": [], |
| "last": "Elmasri", |
| "suffix": "" |
| }, |
| { |
| "first": "B", |
| "middle": [], |
| "last": "Shamkant", |
| "suffix": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Navathe", |
| "suffix": "" |
| } |
| ], |
| "year": 1994, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Ramez Elmasri and Shamkant B. Navathe. Fundamentals of Database Systems, 2nd edition, Reading, MA: Addison-Wesley, 1994.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "Database Systems: Design, Implementation, and Management", |
| "authors": [ |
| { |
| "first": "Peter", |
| "middle": [], |
| "last": "Rob", |
| "suffix": "" |
| }, |
| { |
| "first": "Carlos", |
| "middle": [], |
| "last": "Coronel", |
| "suffix": "" |
| } |
| ], |
| "year": 1997, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Peter Rob, and Carlos Coronel. Database Systems: Design, Implementation, and Management, New York: Course Technology, a Division of International Thomason Publishing, 1997.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "Computing Information by Equation Solving", |
| "authors": [ |
| { |
| "first": "Kiyong", |
| "middle": [], |
| "last": "Lee", |
| "suffix": "" |
| } |
| ], |
| "year": 1993, |
| "venue": "Language, Information, and Computation", |
| "volume": "", |
| "issue": "", |
| "pages": "14--26", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kiyong Lee. \"Computing Information by Equation Solving\", in Chungmin Lee and Beom-mo Kang (eds.), Language, Information, and Computation, pp. 14-26, Seoul: Taehaksa, 1993.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "TABREF0": { |
| "html": null, |
| "text": "The name of this table is PHONE BOOK. The table has four attributes: PROF NO, NAME, INITIAL, and OFFICE_PHONE. Under each attribute, 4 different values are given with each of the four rows forming a record. The first row, for instance, gives information on Lee's office phone number. Tables in an RDBMS are characterized by their data and structural independence. The above table (4), for instance, can easily be expanded by adding new data without destroying its basic structure.Here, the attribute PROF_NO on each of the two tables (4), (5) and (6) uniquely identifies each row, thus playing a role as a primary key for linking one table to the other or joining them into a new expanded table.The notion of table is very simple as sketched here. It is purely a conceptual construct. This paper will, however, show the expressive power of tables for linguistic description. For this purpose, the paper aims at showing how seemingly complex-looking linguistic objects like Phrase Structure trees and logical formulas can be converted into table forms.'5. FROM TREES TO TABLESOne possible way of representing the constituent structure of a sentence is to use a tree form. The following sentence (7), for example, may be analyzed into a phrase structure tree (8):", |
| "content": "<table><tr><td>(4</td><td/><td/><td/><td/></tr><tr><td colspan=\"6\">PROF_NO NAME INITIAL OFFICE_PHONE</td></tr><tr><td>01</td><td/><td colspan=\"2\">Lee</td><td colspan=\"2\">K.</td><td>3290-2171</td></tr><tr><td colspan=\"2\">02</td><td colspan=\"2\">Choe</td><td>J.</td><td>3290-2172</td></tr><tr><td>03</td><td/><td colspan=\"2\">Kang</td><td colspan=\"2\">B.</td><td>3290-2173</td></tr><tr><td colspan=\"2\">04</td><td colspan=\"2\">Kang</td><td colspan=\"2\">M.</td><td>3290-2174</td></tr><tr><td colspan=\"3\">PROF_NO</td><td colspan=\"2\">NAME</td><td>INITIAL</td><td>OFFICE_PHONE</td></tr><tr><td/><td>01</td><td/><td>Lee</td><td/><td>K.</td><td>3290-2171</td></tr><tr><td>_</td><td>02</td><td/><td>Choe</td><td/><td>J.</td><td>3290-2172</td></tr><tr><td/><td>03</td><td/><td>Kang</td><td/><td>B.</td><td>3290-2173</td></tr><tr><td/><td>04</td><td/><td>Kang</td><td/><td>M.</td><td>3290-2174</td></tr><tr><td/><td>05</td><td/><td>Kim</td><td/><td>S.</td><td>3290-2175</td></tr><tr><td/><td>06</td><td/><td>You</td><td/><td>S.</td><td>3290-2176</td></tr><tr><td colspan=\"6\">This new table can be considered as being obtained by joining table (4) with table (6) containing</td></tr><tr><td colspan=\"5\">information on Kim and You.</td></tr><tr><td colspan=\"6\">I PROF_NO NAME INITIAL OFFICE_PHONE</td></tr><tr><td colspan=\"2\">05</td><td/><td>Kim</td><td/><td>S.</td><td>3290-2175</td></tr><tr><td colspan=\"2\">06</td><td/><td>You</td><td/><td>S.</td><td>3290-2176</td></tr><tr><td colspan=\"3\">(7) Mia loves Kim.</td><td/><td/></tr><tr><td>(8)</td><td/><td/><td/><td/></tr><tr><td colspan=\"2\">NP:O1</td><td/><td colspan=\"2\">VP:02</td></tr><tr><td colspan=\"2\">N:011</td><td colspan=\"2\">V:021</td><td colspan=\"2\">NP:022</td></tr><tr><td/><td/><td/><td/><td colspan=\"2\">N:0221</td></tr><tr><td>Mia</td><td/><td colspan=\"2\">loves</td><td colspan=\"2\">Kim</td></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF1": { |
| "html": null, |
| "text": "", |
| "content": "<table><tr><td>: TREE8</td><td/><td/><td/></tr><tr><td>NODE_CODE</td><td>MOTHER</td><td>DAUGHTER1</td><td>DAUGHTER2</td></tr><tr><td>0</td><td>S</td><td>NP</td><td>VP</td></tr><tr><td>01</td><td>NP</td><td>N</td><td/></tr><tr><td>02</td><td>VP</td><td>V</td><td>NP</td></tr><tr><td>011</td><td>N</td><td>t</td><td/></tr><tr><td>021</td><td>V</td><td>t</td><td/></tr><tr><td>022</td><td>NP</td><td>N</td><td/></tr><tr><td>0221</td><td>N</td><td>t</td><td/></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF2": { |
| "html": null, |
| "text": "Here, we first expand the bridging table (11) by adding a few more possible links to it.", |
| "content": "<table><tr><td colspan=\"4\">(13) Table Name: BRIDGING FOR GENERATION</td></tr><tr><td colspan=\"2\">NODE_CODE</td><td>LEX_CODE</td><td/></tr><tr><td>011</td><td/><td>n_l</td><td/></tr><tr><td>011</td><td/><td>n_2</td><td/></tr><tr><td>021</td><td/><td>v_i</td><td/></tr><tr><td>021</td><td/><td>v2</td><td/></tr><tr><td>0221</td><td/><td>n_l</td><td/></tr><tr><td>0221</td><td/><td>n_2</td><td/></tr><tr><td colspan=\"4\">On the basis of this bridging table, we can now complete the table frame (12) and obtain the following:</td></tr><tr><td>S_CODE</td><td>N</td><td>V</td><td>N</td></tr><tr><td>1</td><td>Kim</td><td>hates</td><td>Mia</td></tr><tr><td>2</td><td>Mia</td><td>hates</td><td>Mia</td></tr><tr><td>3</td><td>Kim</td><td>loves</td><td>Kim</td></tr><tr><td>4</td><td>Mia</td><td>loves</td><td>Kim</td></tr><tr><td>5</td><td>Kim</td><td>hates</td><td>Mia</td></tr><tr><td>6</td><td>Mia</td><td>hates</td><td>Mia</td></tr><tr><td>7</td><td>Kim</td><td>loves</td><td>Kim</td></tr><tr><td>8</td><td>Mia</td><td>loves</td><td>Kim</td></tr><tr><td>(10)</td><td/><td/><td/></tr><tr><td colspan=\"3\">: PARSING8</td><td/></tr><tr><td colspan=\"3\">S_CODE I N I V I N</td><td/></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF3": { |
| "html": null, |
| "text": "table consisting of four attributes, PROP_CODE, RELATION, ARG1, and ARG2.", |
| "content": "<table><tr><td>PROP_CODE</td><td>RELATION</td><td>ARG1</td><td>ARG2</td></tr><tr><td>1</td><td>love</td><td>mia</td><td>kim</td></tr><tr><td colspan=\"4\">The first row in this table is marked with PROP_CODE 1, containing information on the loving relation</td></tr><tr><td>between Mia and Kim.</td><td/><td/><td/></tr><tr><td colspan=\"3\">To this table, other relations may be added.</td><td/></tr><tr><td>PROP CODE</td><td>RELATION</td><td>ARG1</td><td>ARG2</td></tr><tr><td>1</td><td>love</td><td>mia</td><td>kim</td></tr><tr><td>2</td><td>love</td><td>kim</td><td>mia</td></tr><tr><td>3</td><td>sisters</td><td>mia</td><td>kim</td></tr><tr><td colspan=\"4\">Two more records or pieces of information are added to the table without changing its basic structural</td></tr><tr><td>frame.</td><td/><td/><td/></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF4": { |
| "html": null, |
| "text": "", |
| "content": "<table><tr><td>18) is thus understood as representing the propositional content of both of the sentences in (19) as</td></tr><tr><td>well.</td></tr><tr><td>6.2 EMBEDDED SENTENCES</td></tr><tr><td>Sentence (20) contains an embedded sentence.</td></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF5": { |
| "html": null, |
| "text": "Here the equation x=mia' is treated as part of the propositional content while its PROP_CODE is assigned prop_4 in the table PROP_(23). But the information that is conveyed by this equation is not part of the propositional content of sentence (20). It is rather part of the background for interpreting it. Hence, it should be taken away from the table PROP_(23) and placed in a separate table for storing background information.As is revised as in (26), thePROP table can link to the BACKGROUND table.", |
| "content": "<table><tr><td colspan=\"2\">: PROP_(23)</td><td/><td/><td/></tr><tr><td>PROP_CODE</td><td>RELATION</td><td>ARG 1</td><td>ARG2</td><td/></tr><tr><td>prop_ 1</td><td>knows'</td><td>mia'</td><td>p</td><td/></tr><tr><td>prop_2</td><td>=</td><td>p</td><td>prop_3</td><td/></tr><tr><td>prop_3</td><td>loves'</td><td>kim'</td><td>x</td><td/></tr><tr><td>prop_4</td><td>=</td><td>x</td><td>mia'</td><td/></tr><tr><td colspan=\"2\">(25) Table name: BACKGROUND</td><td/><td/><td/></tr><tr><td>BG_CODE</td><td>RELATION</td><td>ARG 1</td><td>ARG2</td><td/></tr><tr><td>bg_l</td><td/><td>x</td><td>mia'</td><td/></tr><tr><td>PROP_CODE</td><td>RELATION</td><td>ARG 1</td><td>ARG2</td><td>BG_CODE</td></tr><tr><td>prop_l</td><td>knows'</td><td>mia'</td><td>p</td><td/></tr><tr><td>prop_2</td><td>=</td><td>p</td><td>prop_3</td><td/></tr><tr><td>prop_3</td><td>loves'</td><td>kim'</td><td>x</td><td>bg_l</td></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF6": { |
| "html": null, |
| "text": ") is now linked to table (25) by relating the PROP_CODE prop_3 to the BG_CODE bg_1.The table approach can distribute information into appropriately classified tables. But it can also link them systematically, as the need arises. Consider another type of embedded sentences:", |
| "content": "<table/>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF7": { |
| "html": null, |
| "text": "table form, we construct a PROP table first.", |
| "content": "<table><tr><td>PROP_CODE</td><td>TENSE</td><td>RELATION</td><td>ARG 1</td><td>ARG2</td><td>ARG3</td><td colspan=\"2\">BG_CODE CL_CODE</td></tr><tr><td>Prop_l</td><td>past</td><td>promise</td><td>mia</td><td>kim</td><td>prop2</td><td/></tr><tr><td>Prop_2</td><td/><td>help</td><td>x</td><td>y</td><td/><td>bg_l 0</td><td>subject</td></tr><tr><td colspan=\"7\">We then need two tables, a table for BACKGROUND and a table for CONTROL.</td></tr><tr><td>BG_CODE</td><td>RELATION</td><td>ARG 1</td><td>ARG2</td><td/><td/><td/></tr><tr><td>Bg_ 1 0</td><td/><td>y</td><td>kim'</td><td/><td/><td/></tr><tr><td>CL_CODE</td><td>RELATION</td><td>ARG I</td><td>ARG2</td><td/><td/><td/></tr><tr><td>Subject</td><td/><td>x</td><td>mia'</td><td/><td/><td/></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF8": { |
| "html": null, |
| "text": "table requires an expanded table for BACKGROUND.", |
| "content": "<table><tr><td>(38)</td><td/><td/><td/></tr><tr><td>BG_CODE</td><td>RELATION</td><td>ARG 1</td><td>ARG2</td></tr><tr><td>bg_9</td><td/><td>y</td><td>mia'</td></tr><tr><td>bg_ 10</td><td>=</td><td>y</td><td>kim'</td></tr></table>", |
| "type_str": "table", |
| "num": null |
| }, |
| "TABREF9": { |
| "html": null, |
| "text": "The PROP table contains three propositions, propi, prop_2, and prop_3. The first row prop_i simply states that the main event of promising occurred in the PAST. The second row represents the main content of sentence (33), but a part of it refers to the third row prop_3. Furthermore, it contains information on T_CODE. Its value temp_2 then relates the PROP table to the TEMP_REL table that tells how the two events are temporally related to each other.", |
| "content": "<table><tr><td/><td/><td/><td/><td/><td/><td/><td>table will contain the</td></tr><tr><td colspan=\"2\">following set of data:</td><td/><td/><td/><td/><td/></tr><tr><td>(36)</td><td/><td/><td/><td/><td/><td/></tr><tr><td>P_CODE</td><td>REL</td><td colspan=\"2\">EVENT ARG 1</td><td>ARG2</td><td>ARG3</td><td colspan=\"2\">BG_CODE CL CODE T_CODE</td></tr><tr><td>prop_ 1</td><td>PAST</td><td>e 1</td><td/><td/><td/><td/></tr><tr><td>prop_2</td><td>promise'</td><td>e I</td><td>bill'</td><td>mia'</td><td>prop_3</td><td/><td>temp_2</td></tr><tr><td>prop_3</td><td>marry'</td><td>e2</td><td>x</td><td>y</td><td/><td>bg_9</td><td>equi-np</td></tr><tr><td>T_CODE</td><td colspan=\"2\">REL</td><td>ARG 1</td><td>ARG2</td><td/><td/></tr><tr><td>temp_ 1</td><td colspan=\"2\">overlaps</td><td>e 1</td><td>e2</td><td/><td/></tr><tr><td>temp_2</td><td colspan=\"2\">precedes</td><td>e 1</td><td>e2</td><td/><td/></tr><tr><td>temp_3</td><td colspan=\"2\">follows</td><td>e2</td><td>e1</td><td/><td/></tr></table>", |
| "type_str": "table", |
| "num": null |
| } |
| } |
| } |
| } |
