An approach of intelligent searching of information in texts

Abstract

Paper proposes an approach aimed at question oriented searching of information in texts. Texts are parsed, keywords and extra features of questions are marked, and sentences in text with the most relevant information to question are defined. Proposed approach is applied to Cyrillic and Latin languages. Case study illustrates how to obtain answers to questions about Bulgarian fairytale that is represented on different languages (Bulgarian and English). Evaluation of the proposed approach is introduced. Description of the software architecture and source code of the corresponding software system are represented. Data structures and examples of *.xml files for storing information about question and answers are outlined.