This paper is devoted to the research and development of machine learning methods aimed at discovering potentially dangerous extremist information in social networks using pattern based approach. In this approach, a text document containing extremist information is used for automatic extracting keywords to query a social networks search engine, and then found messages are filtered according to the topic based measure of relevance with the pattern. N-gram based algorithms are proposed for constructing hidden topics and keywords that allow applying the proposed approach in the case of multilingual and illiterate texts. The performance of the proposed methods is experimentally studied on benchmark Ansar1 dataset.
CITATION STYLE
Petrovskiy, M., Tsarev, D., & Pospelova, I. (2017). Pattern based information retrieval approach to discover extremist information on the internet. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10682 LNAI, pp. 240–249). Springer Verlag. https://doi.org/10.1007/978-3-319-71928-3_24
Mendeley helps you to discover research relevant for your work.