Abstract: Fact extraction from the text is one of the most important areas of Natural Language Processing (NLP). Majority of existing approaches allows extracting facts from structured textual information of the specific subject areas. This paper proposes a logical-linguistic model extracting facts from semi-structured texts in English, which belong to unlimited subject areas. A fact is written in the form of a triplet: Subject - Predicate - Object, in which the Predicate defines the relations and Subject and Object define the subjects, objects or concepts. Our model defines meaning relations via grammatical and semantic features of the words in English sentences. In order to formalize and represent the participants of the fact triplet explicitly, we identify subject variables. The subject variables define a finite set of morphological and syntactic features of the words in sentences. The model was successfully implemented in the system of extraction and identification of a few types of the facts: the fact of lacking, the fact of ownership, the fact of transferring, and the fact of the presence of the attribute of time, location, and belonging for the first three fact actions. We estimated the effectiveness of our model via the coefficients of precision and recall. Results оf the paper show that using of the model lets increase the numerical values of these coefficients.

Keywords: facts extraction from the text, Natural Language Processing, semantic relations, the algebra of finite predicates, recall and precision.

ACM Classification Keywords: H.3.3 .Information Search and Retrieval, I.2.4. Knowledge Representation Formalisms and Methods

Link:

Facts extraction from the semi-structured text information

Nina Khairova, Nataliya Sharonova, Ajit Pratap Singh Gautam

http://www.foibg.com/ijima//vol05/ijima05-01-p07.pdf