Лингвистические инструменты выявления криминально окрашенной текстовой информации веб-контента

Abstract

The paper deals with types of criminally colored textual information of Web content (cybercrime, terrorist act or financial fraud) and analyzes existing technologies of linguistic analysis that allow to identify illegal information in texts. An analytical review of the use of existing language processing tools is conducted, which allows to identify problems of using traditional NLP approaches for the analysis of criminal-significant textual information. The proposed approach is based on the approaches of Information Extraction and focuses on the method of extracting facts from weakly structured texts. The use of technology based on the description of semantic functions by means of algebra of finite predicates is considered, to extract weakly structured facts from sentences of Russian and English. The possibility of using the proposed technology for the analysis of the Kazakh language texts is analyzed.