1. MAKING ONTOLOGIES WORK for RESOLVING REDUNDANCIES ACROSS DOCUMENTS.
- Author
-
Everett, John O., Bobrow, Daniel G., Stolle, Reinhard, Crouch, Richard, De Paiva, Valeria, Condoravdi, Cleo, Van den Berg, Martin, and Polanyi, Livia
- Subjects
- *
ONTOLOGY , *REALIZATION (Linguistics) , *SEMANTICS (Philosophy) , *CONCEPTS , *IDEA (Philosophy) - Abstract
This article focuses on ways to produce normalized representations in ontology from a wide range of different ways of expressing the same idea and describes a particular mechanism for normalizing frequently occurring comparative constructions. To create useful representations of natural language text, first obtain a compact representation of the syntactic and semantic structures for each sentence, using the Xerox Linguistic Environment, a deep parser based on Lexical Functional Grammar theory. Reasoning systems that receive well-specified input can utilize carefully constrained ontologies that capture exactly the set of concepts necessary for the task at hand. Matching based on midlevel concepts and role comparisons is one mechanism for assessing similarity, but others need to handle some common linguistic constructions. Two design criteria have been discussed for ontologies to support the task of finding similarities and redundancies in an ontology to support tasks such as identification of the appropriate level of abstraction in representation, and the normalization of dimensions for comparatives. In general, it is balancing adequacy in expressiveness against complexity in similarity reasoning. As of February 2002, the system normalizes the dimensional comparisons that occur in the test set of 15 similar pairs of documents and exploits mid-level categories to make similarity assessments. It is expected that these criteria, along with others that will emerge as the research progresses, will form the basis for defining a powerful yet tractable ontology for large-scale knowledge extraction from documents.
- Published
- 2002
- Full Text
- View/download PDF