SUMO mapping to wordnets 

Context in source publication

Context 1

... given an English synset, all corresponding Arabic variants (if any) will be selected; given an Arabic word, all its senses are determined and for each of them the corresponding English synset is encoded. The Arabic synsets will be extended with hypernym relations to form a closed semantic hierarchy. SUMO will be used to maximize the semantic consistency of the hyponymy links. This will represent the core wordnet, which is a semantic basic for the further extension. The work is mostly done manually. When a new Arabic verb is added, extensions are made from verbal entries, including verbal derivates, nominalizations, verbal nouns, and so on. We also consider the most productive forms of deriving broken plurals. This is done by applying lexical and morphological rules iteratively. The database is further extended downward from the CBCs. First, a layer of hyponyms is chosen based on maximal connectivity, relevance, and generality. Two major pre-processing steps are required, preparation and extension. Preparation entails compiling lexical and morphological rules and processing available bilingual resources from which we construct a homogeneous bilingual dictionary containing information on the Arabic/English word pair. This information includes the Arabic root, the POS, the relative frequencies and the sources supporting the pairing. The Arabic words in these bilingual resources must also be normalized and lemmatized while maintaining vowels and diacritics. We next apply 17 heuristic procedures, previously used for EWN, to the bilingual dictionary in order to derive candidate Arabic words/English synsets mappings. Each mapping includes the Arabic word and root, the English synset, the POS, the relative frequencies, a mapping score, the absolute depth in AWN, the number of gaps between the synset and the top of the AWN hierarchy, and attested tokens of the pair. The Arabic word/English synset pairs constitute the input to a manual validation process. We proceed by chunks of related units (sets of related WN synsets, e.g. hyponymy chains and sets of related Arabic words, i.e., words having the same root) instead of individual units (i.e., synsets, senses, words). Finally, AWN will be completed by filling in the gaps in its structure, covering specific domains, adding terminology and named entities, etc. Each synset construction step is followed by a validation phase, where formal consistency is checked and the coverage is evaluated in terms of frequency of occurrence and domain distribution. The total coverage of AWN will be around 10,000 synsets. Tools to be developed for AWN include a lexicographer's interface modeled on the EWN interface with added facilities for Arabic script. Because AWN is to be aligned not just to PWN but to every wordnet aligned to PWN – either directly or indirectly through an Interlingual Index or the ontology – the database design supports multiple languages. The user interface will be explicitly multilingual and indifferent to the direction of alignment between the conceptual structures of the two languages. In addition to search and browsing facilities for the end users of the completed database, lexicographers require an editing interface. A variety of legacy components are available, each with their relative advantages. The editor's interface will communicate with the database server using Simple Object Access Protocol (SOAP), allowing multiple lexicographers at different sites to maintain a common database. The database structure comprises four principal entity types: item , word , form and link . Items are conceptual entities, including synsets, ontology classes and instances. An item has a unique identifier and descriptive information such as a gloss. Items lexicalized in different languages are distinct. A word entity is a word sense, where the word's citation form is associated with an item via its identifier. A form is an entity that contains lexical information (not merely inflectional variation). The forms are the root and/or the broken plural form, where applicable. A link relates two items, and has a type such as "equivalence," "subsuming," etc. Links interconnect sense items, e.g., a PWN synset to an AWN synset, a synset to a SUMO concept, etc. This data model has been specified in XML as an interchange format, but is also implemented in a MySQL database hosted by one of the partners. A large ontology providing the semantic underpinning for AWN concepts will be built on SUMO, a formal ontology of about 1000 terms and 4000 definitional statements currently that is provided in a first order logic language called Standard Upper Ontology Knowledge Interchange format (SUO-KIF) and also translated into OWL semantic web language. SUMO has natural language generation templates and a multi-lingual lexicon that allows statements in SUO-KIF and SUMO to be expressed in multiple languages. Synsets map to a general SUMO term or a term that is directly equivalent to the given synset (Figure 1). New formal terms will be defined to cover a greater number of equivalence mappings, and the definitions of the new terms will in turn depend upon existing fundamental concepts in SUMO. The process of formalizing definitions will generate feedback as to whether word senses in AWN need to be divided or combined and how glosses may be clarified. Wordnets in other languages linked by synset number will benefit, too. The Sigma ontology development environment will be updated to handle a similar presentation of Unicode-based character sets, including Arabic. The Interlingual Index (ILI) connecting EWN wordnets is a condensed set of more or less universal concepts linking synsets across languages via multiple exhaustive equivalence relations. In EuroWordNet and BalkaNet, English PWN has been used to express equivalence relations across the different languages. By providing many SUMO definitions and terms that correspond to Arabic synsets, we will create the opportunity to use SUMO as the ILI for all wordnets that are currently related to PWN. This is illustrated in Figure 2. If the Arabic word sense for shai is exhaustively defined by relations to SUMO terms, this definition can replace an equivalence relation (er1) that is currently encoded between the Arabic synset shai and a synset tea in PWN. Note that the relations from shai to the SUMO terms need to be exhaustive, which may require multiple relations of different types (sr1 (subsumption), r2, r3) to multiple SUMO terms. If there are also equivalence relations from other languages (e.g. Dutch and Spanish) to the same PWN synset, then these relations grant the linkage of the synsets in these languages to the same SUMO definition. Besides providing a formal semantic framework, SUMO can thus also be used to map synsets across languages, in fact even when there is not an equivalent in English. By composing formal definitions for the non-English synsets, SUMO as an ILI will not only be less biased by English but also has more expressive power. Constructing AWN presents challenges not encountered by established wordnets. These include the script on the one hand and the morphological properties of Semitic languages, centered around roots, on the other hand. The foundations for meeting these challenges have been laid. An innovation with significant consequences for wordnet development is the proposal to substitute English WN as the ILI with SUMO. This work was supported by the United States Central Intelligence Agency. Beesley, K. (2001) Finite-State Morphological Analysis and Generation of Arabic at Xerox, ACL/EACL 2001, July 6 th , Toulouse, France : 1-8 Black, W., Elkateb, S., Rodriguez, H, Alkhalifa, M., Vossen, P., Pease, A. and Fellbaum, C., (2006). Introducing the Arabic WordNet Project, in Proceedings of the Third International WordNet Conference, Sojka, Choi, Fellbaum and Vossen eds. Black, W. J., and Elkateb, S. (2004) A Prototype English- Arabic Dictionary Based on WordNet, Proceedings of 2 nd Global WordNet Conference, GWC2004, Czech Republic, 67-74. Buckwalter, T. (2002) Arabic Morphological Analysis, Http://www.qamus.org/morphology.htm De Roeck, A., and Al-Fares, W. (2000) A Morphologically Sensitive Clustering Algorithm for Identifying Arabic Roots Proceedings of the 38 th Annual Meeting of the ACL, Hong Kong, 199-206 Dyvik, H. (2003) Translations as a semantic knowledge source: word alignment and wordnet, Section for Linguistic Studies scientific papers, University of Bergen Dyvik, H. (2002) Translations as Semantic Mirrors: From Parallel Corpus to Wordnet1. Section for Linguistic Studies scientific papers, University of Bergen Elkateb, S and Black, W. J. (2001) Towards the Design of English-Arabic Terminological Knowledge Base, Proceedings of ACL 2000, Toulouse, France:113-118 Elkateb, S and Black, W. J. (2004) A Bilingual Dictionary with Enriched Lexical Information, Proceedings of NEMLAR Cairo, Egypt 2004 Arabic Language Tools and Resources: 79-84 Elkateb, S. (2005) Design and implementation of an English Arabic dictionary/editor. PhD thesis, The University of Manchester, United Kingdom. Farreres, J. (2005) Creation of wide-coverage domain- independent ontologies. PhD thesis, Univertitat Politècnicade Catalunya. Fellbaum, C., (1998, ed.) WordNet: An Electronic Lexical Database. Cambridge, MA: MIT Press . Niles, I., and Pease, A. (2001) Towards a Standard Upper Ontology. In: Proceedings of FOIS 2001, Ogunquit, Maine, pp. 2-9. Pease, A., (2000) Standard Upper Ontology Knowledge Interchange Format. Web document Pease, A., (2003) The Sigma Ontology Development Environment, in Working Notes of the IJCAI-2003 Workshop on Ontology and Distributed Systems. Volume 71 of CEUR Workshop Proceeding series Pustejovsky, J. (1995) The Generative Lexicon, Massachusetts Institute of Technology. Tufis, D. (ed.) (2004) Special Issue on the BalkaNet project. Romanian Journal of Information Science and Technology, Vol.7, nos 1-2 Vossen, P. (ed.) (1998) ...