Corereference or referential identity - the relationship between names - the components of the utterance, in which the names refer to the same object (situation) of extra-linguistic reality ( referent ) [1] .
Due to the coherence, the text can be made coherent, often, but not necessarily, in the form of an anaphoric relationship (pronoun or meaning of certainty - pronoun ).
Content
Examples
Anaphoric relation “pronoun word - pronoun”: “ The book lies on the table. She’s heavy. ”
Anaphorical relation “the meaning of certainty is a pronoun”: “ The book lies on the table. Tom is heavy. ”
Metonymy : “ The writer forced the public to read himself .” (Here himself - the writer's works)
Coreference of names in computer information processing
The name in information systems with a tabular form for presenting information ( databases , web sites , spreadsheets , etc.) is represented by an attribute or a set of attributes of a tuple (for name groups). Symbolically different names in such systems can be referential, and attributes containing them can be semantically equivalent . For example, the name “cleaning agent” is coherent with the name “cleaning agent” . [2]
Undoubtedly, in the absence of special rules, reference names will be entered by the operators of different information systems symbolically different. Usually, the task of determining the semantic equivalence of symbolic attributes is considered algorithmically unsolvable. At the level of individual software solutions, states and the world community as a whole apply technologies that allow you to implicitly transfer the name comparison function to a computer operator, and present the result as a code, or in another algorithmically convenient form.
- In the Russian Federation, each taxpayer must have its own TIN
- Each product in any supermarket in the world has a unique numeric code encoded in a bar code.
- GOST 7.1-2003 introduces a single standard for bibliographic records and bibliographic descriptions
- Dublin Core is the standard for describing the widest range of network resources.
- CommerceML uses various formats of electronic document exchange
Advances in computer analysis of natural languages make it possible to create information systems for solving such problems where the introduction of code marks or naming standards is impossible. The pioneer in Runet for automatic comparison of semantics of names can be considered the Price.ru system for searching for information about products. It uses the methods of information retrieval systems, which allows you to establish the semantic tolerance of various items.
Most of the research and publications are related to the automation of the establishment of coreferential relationships in a natural language text. The use of domain-specific semantic models allows you to effectively perform semantic comparisons of different names character-by-character in databases for various application purposes.
Notes
- ↑ Linguistic Encyclopedic Dictionary , “Coreference”
- ↑ Ahmed K. Elmagarmid, Panagiotis G. Ipeirotis, Vassilios S. Verykios. Duplicate Record Detection: A Survey . IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 19, NO. 1, JANUARY 2007. Archived July 22, 2012.
Literature
- Linguistic Encyclopedic Dictionary
Links
- Glossary on the website of the Expert Forensicist, “Coreference (reference identity”)
- Eric Bengtson, Dan Roth Understanding the Value of Features for Coreference Resolution
- Ermakov A. E. Reference designations of persons and organizations in Russian-language media texts: empirical laws for computer analysis
- Berdnik V. L. Models and methods of semantic comparison of character strings in a collection of documents