| { |
| "paper_id": "E99-1040", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T10:37:26.039219Z" |
| }, |
| "title": "Designing spelling correctors for inflected languages using lexical transducers", |
| "authors": [ |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Aldezabal", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of the Basque Country", |
| "location": { |
| "addrLine": "649 postakutxa", |
| "postCode": "20080" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Alegria", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of the Basque Country", |
| "location": { |
| "addrLine": "649 postakutxa", |
| "postCode": "20080" |
| } |
| }, |
| "email": "i.alegria@si.ehu.es" |
| }, |
| { |
| "first": "O", |
| "middle": [], |
| "last": "Ansa", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of the Basque Country", |
| "location": { |
| "addrLine": "649 postakutxa", |
| "postCode": "20080" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "J", |
| "middle": [ |
| "M" |
| ], |
| "last": "Arriola", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of the Basque Country", |
| "location": { |
| "addrLine": "649 postakutxa", |
| "postCode": "20080" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Ezeiza", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "University of the Basque Country", |
| "location": { |
| "addrLine": "649 postakutxa", |
| "postCode": "20080" |
| } |
| }, |
| "email": "" |
| }, |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Aduriz", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "UZEI Hizkia", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "A", |
| "middle": [], |
| "last": "Da Costa", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "UZEI Hizkia", |
| "location": {} |
| }, |
| "email": "" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "", |
| "pdf_parse": { |
| "paper_id": "E99-1040", |
| "_pdf_hash": "", |
| "abstract": [], |
| "body_text": [ |
| { |
| "text": "This paper describes the components used in the design of the commercial XuxenII spelling checker/corrector for Basque. It is a new version of the Xuxen spelling corrector (Aduriz et al., 97) which uses lexical transducers to improve the process. A very important new feature is the use of user dictionaries whose entries can recognise both the original and inflected forms. In languages with a high level of inflection such as Basque spelling checking cannot be resolved without adequate treatment of words from a morphological standpoint. In addition to this, the morphological treatment has other important features: coverage, reusability of tools, orthogonality and security. The tool is based in lexical transducers and is built using the fst library of Inxight 1. A lexical transducer (Karttunen, 94 ) is a finite-state automaton that maps inflected surface forms to lexical forms, and can be seen as an evolution of twolevel morphology (Koskenniemi, 83) where the use of diacritics and homographs can be avoided and the intersection and composition of transducers is possible. In addition, the process is very fast and the transducer for the whole morphological description can be compacted in less than 1Mbyte. The design of the spelling corrector consists of four main modules:", |
| "cite_spans": [ |
| { |
| "start": 172, |
| "end": 187, |
| "text": "(Aduriz et al.,", |
| "ref_id": null |
| }, |
| { |
| "start": 188, |
| "end": 191, |
| "text": "97)", |
| "ref_id": null |
| }, |
| { |
| "start": 791, |
| "end": 802, |
| "text": "(Karttunen,", |
| "ref_id": null |
| }, |
| { |
| "start": 803, |
| "end": 805, |
| "text": "94", |
| "ref_id": null |
| }, |
| { |
| "start": 943, |
| "end": 956, |
| "text": "(Koskenniemi,", |
| "ref_id": null |
| }, |
| { |
| "start": 957, |
| "end": 960, |
| "text": "83)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "\u2022 the standard checker,", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "\u2022 the recogniser using user-lexicons,", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "\u2022 the corrector of linguistic variants -proposals for dialectal uses and competence errors-. the corrector of typographical errors An important feature is its homogeneity. The different steps are based on lexical transducers, far from ad-hoc solutions.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "lInxight Software, Inc., a Xerox New Enterprise Company (www.inxight.com)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Introduction", |
| "sec_num": "1" |
| }, |
| { |
| "text": "The spelling checker accepts as correct any word which allows a correct standard morphological breakdown. When a word is not recognised by the checker, it is assumed to be a misspelling and a warning is given to the user who has different options, being one of most interesting including its lemma in the user-lexicon.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The Spelling Checker", |
| "sec_num": "2" |
| }, |
| { |
| "text": "The user-lexicon is offered in order to increase the coverage and to manage specific terminology. Our tool recognises all the possible inflections of a root. The use of a lexical transducer for this purpose is difficult because it is necessary to compile the new entries with the affixes and the rules to update it but this process is slow. The mechanism we have implemented has the following two main components in order to be able to treatment declensions:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The user lexicons", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "1. a general transducer which use standard rules but totally opened lexicon. The result of the analysis is not only if the word is known or not, but also all the possible lemmas corresponding to this word-form and the grammatical category of each one. The resulting lexical transducer is very compact and fast.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The user lexicons", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "2. a searcher of these hypothetical lemmas in the user-lexicons. If one of them is found, the checker will accept the word, otherwise it will suppose that it has to be corrected.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The user lexicons", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "For this process the system has an interface to update the user lexicon because the part of speech of the lemmas is necessary when they are added to the user lexicon.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The user lexicons", |
| "sec_num": "2.1" |
| }, |
| { |
| "text": "The Spelling Corrector", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "3", |
| "sec_num": null |
| }, |
| { |
| "text": "Although there is a wide bibliography about the problem of correction (Kukich, 92) , it is significative that almost all of them do not mention the relation with morphology and assume that there is a whole dictionary of words or that the system works without lexical information. Oflazer and Guzey (1994) face the problem of correcting words in agglutinative languages.", |
| "cite_spans": [ |
| { |
| "start": 70, |
| "end": 78, |
| "text": "(Kukich,", |
| "ref_id": null |
| }, |
| { |
| "start": 79, |
| "end": 82, |
| "text": "92)", |
| "ref_id": null |
| }, |
| { |
| "start": 280, |
| "end": 304, |
| "text": "Oflazer and Guzey (1994)", |
| "ref_id": "BIBREF5" |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "3", |
| "sec_num": null |
| }, |
| { |
| "text": "The need of managing competence errors -also named orthographic errors-has been mentioned and reasoned by different authors (van Berkel &: de Smedt, 88). When we faced the problem of correcting misspelled words the main problem found was that because of the recent standardisation and the widespread dialectal use of Basque, competence errors or linguistic variants are more likely and therefore their treatment becomes critical. When we decided to use lexical transducers for the treatment of linguistic variants, the following procedure was applied to build the transducer:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "1. Additional morphemes are linked to the standard ones using the possibility of expressing two levels in the lexicon.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "2. Definition of additional rules for competence errors that do not need to be integrated with the standard ones. It is possible and clearer to put these rules in other plane near to the surface and compose them with the standard rules, because most of the additional rules are due to phonetic changes.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "When a word-form is not accepted the word is checked against this second transducer. If the incorrect form is recognised now -i.e. it contains a competence error-the correct lexical level form is directly obtained and, as the transducers are bi-directional, the corrected surface form will be generated from the lexical form using only standard transducer.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "For example, the word-form beartzetikan, misspelling of behartzetik (from the need) can be corrected although the edit-distance is three. The process of correction is the following:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "\u2022 Decomposition into three morphemes: behar (using a rule to guess the h), tze and tikan.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "\u2022 tikan is a non-standard use of tik and as they are linked in the lexicon is chosen.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "* The standard generation of behar+tze+tik obtains the correct word behartzetik.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Correcting Competence Errors", |
| "sec_num": "3.1" |
| }, |
| { |
| "text": "The treatment of typographical errors is quite conventional and performs the following:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Handling Typographical Errors", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "\u2022 Generating proposals to typographical errors using Damerau's classification (edit distance of one). These proposals are ranked in order of trigramic probability.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Handling Typographical Errors", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "\u2022 Spelling checking of proposals.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Handling Typographical Errors", |
| "sec_num": "3.2" |
| }, |
| { |
| "text": "The results are very good in the case of competence errors and not so good for typographical errors because in the last case only errors with an edit-distance of one have been planned. In 89right proposal is generated and in 71possible to generate and test all the possible words with an editdistance higher, but the number of proposal would be very high. The corrector has been integrated in several tools. A demonstration can be seen in http://ixa.si.ehu.es.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Results", |
| "sec_num": "3.3" |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "Acknowledgements This work has had partial support from the Culture Department of the Government of the Basque Country. We would like to thank to Xerox for letting us using their tools, and also to Lauri Karttunen for his help.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "acknowledgement", |
| "sec_num": null |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "A spelling corrector for Basque based on morphology", |
| "authors": [ |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Aduriz", |
| "suffix": "" |
| }, |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Alegria", |
| "suffix": "" |
| }, |
| { |
| "first": "X", |
| "middle": [], |
| "last": "Artola", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Ezeiza", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Sarasola", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Urkia", |
| "suffix": "" |
| } |
| ], |
| "year": 1997, |
| "venue": "Literary & Linguistic Computing", |
| "volume": "12", |
| "issue": "1", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Aduriz I., Alegria I., Artola X., Ezeiza N., Sara- sola K., Urkia M. (1997), A spelling corrector for Basque based on morphology. Literary & Linguistic Computing, Vol. 12, No. 1. Oxford University Press. Oxford.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Improving a Robust Morphological AnaIyser using Lexical Transducers. Recent Advances in Natural Language Processing", |
| "authors": [ |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Alegria", |
| "suffix": "" |
| }, |
| { |
| "first": "X", |
| "middle": [], |
| "last": "Artola", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Sarasola", |
| "suffix": "" |
| } |
| ], |
| "year": 1997, |
| "venue": "Current Issues in Linguistic Theory (CILT) series. John Benjamins publisher company", |
| "volume": "136", |
| "issue": "", |
| "pages": "97--110", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Alegria I., Artola X., Sarasola K (1997). Improv- ing a Robust Morphological AnaIyser using Lex- ical Transducers. Recent Advances in Natural Language Processing. Current Issues in Linguis- tic Theory (CILT) series. John Benjamins pub- lisher company. Vol. 136. pp 97-110.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Constructing Lexical Transducers", |
| "authors": [ |
| { |
| "first": "L", |
| "middle": [], |
| "last": "Karttunen", |
| "suffix": "" |
| } |
| ], |
| "year": 1994, |
| "venue": "Proc. of COLING'94", |
| "volume": "", |
| "issue": "", |
| "pages": "406--411", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Karttunen L. (1994). Constructing Lexical Trans- ducers, Proc. of COLING'94, 406-411.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Two-level Morphology: A general Computational Model for Word-Form Recognition and Production", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Koskenniemi", |
| "suffix": "" |
| } |
| ], |
| "year": 1983, |
| "venue": "Department of General Linguistics. Publications", |
| "volume": "", |
| "issue": "11", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Koskenniemi, K. (1983). Two-level Morphology: A general Computational Model for Word- Form Recognition and Production, University of Helsinki, Department of General Linguistics. Publications No. 11.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Techniques for automatically correcting word in text", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Kukich", |
| "suffix": "" |
| } |
| ], |
| "year": 1992, |
| "venue": "ACM Computing Surveys", |
| "volume": "24", |
| "issue": "4", |
| "pages": "377--439", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Kukich K. (1992). Techniques for automatically correcting word in text. ACM Computing Sur- veys, vol. 24, No. 4, 377-439.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "Spelling Correction in Aglutinative Languages", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Oflazer", |
| "suffix": "" |
| }, |
| { |
| "first": "C", |
| "middle": [], |
| "last": "Guzey", |
| "suffix": "" |
| } |
| ], |
| "year": 1994, |
| "venue": "Proc. of ANLP-94", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Oflazer K, Guzey C. (1994). Spelling Correction in Aglutinative Languages, Proc. of ANLP-94, Sttutgart.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "Triphone analysis: a combined method ]or the correction o] orthographic and typographical errors", |
| "authors": [ |
| { |
| "first": "B", |
| "middle": [], |
| "last": "Van Barkel", |
| "suffix": "" |
| }, |
| { |
| "first": "De", |
| "middle": [], |
| "last": "Smedt", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "", |
| "suffix": "" |
| } |
| ], |
| "year": 1988, |
| "venue": "Procedings of the Second Conference ANLP (ACL)", |
| "volume": "", |
| "issue": "", |
| "pages": "77--83", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Van Barkel B, De Smedt K. (1988). Triphone anal- ysis: a combined method ]or the correction o] orthographic and typographical errors. Proced- ings of the Second Conference ANLP (ACL), pp.77-83.", |
| "links": null |
| } |
| }, |
| "ref_entries": {} |
| } |
| } |
