B.N. Onykiy - Dr. Sc. (Eng.), Professor, Head of Department №65 «Analysis of Competitive Systems», National Research Nuclear University «MEPhI» (Moscow) E-mail: bnonykij@mephi.ru K.A. Sokolina - Assistant, Department №65 «Analysis of Competitive Systems», National Research Nuclear University «MEPhI» (Moscow) E-mail: k.sokolina@mail.ru

The paper is dedicated to the issues of developing multiagent information analysis systems (MIAS) for search and processing of scientific and technical information. The main purpose of such systems is providing regular information services for specialists in various disciplines and those conducting scientific research and engineering activities. With the help of automatically initiated search agents MIAS technologies help to simplify working with large volumes of data scattered over a variety of Internet sources. Agent user service is carried out in accordance with pre-defined search statements including user-s index of thematic interest; objects of professional interest and parameters and conditions describing them; groups of objects of professional interest and types of relations between them. According to their statements, users receive analytical reports on a regular basis. They are news digests on specified topics, text reports (profiles) and semantic networks on user-s objects of interest and groups of objects. The practice of applying an industrial IT platform («Semantic Archive») for agent systems engineering is described in the paper. The structure of the software solution including server and client components is presented. The conceptual model of MIAS for scientific and technological information is described in detail. It covers major functional modules and databases which make up the system-s core. Two databases play a key role in the conceptual MIAS model. They are the thematic routes database (TRDB) and the thesauruses database (TDB), because they in particular are used for agent search management. The structure of the MIAS core is provided. It is the minimum system configuration sufficient to launch regular use and release of analytical reports on a limited number of topics with a capability of further scaling-up by the number of topics and search languages. Two processing circuits (target and service) are distinguished. The target circuit is where regular agent monitoring of thematic groups of sources (with the use of the thematic routes database) and agent collections filtering (with the use of multilingual thematic thesauruses) is carried out. Target circuit also has a link to the user and a link to the text mining channel. Service circuit provides target circuit management. The operator who makes decisions about TRDB updates controls its functioning. The sources - accessibility is described by the traffic lights principle. The operator is notified about «yellow» and «red» sources that require extra actions or tuning. Man-machine procedures of TRDB updates are presented in terms of algorithms. However not only new sources are included, but also inactive sources are archived. Update procedures are conducted dynamically without interrupting the target circuit operation and based on its output. As a result, thematic routes database does not go out of date providing consistently high recall.

 

1. Daniel Reis Cybersecurity: Issues of Today, a Path for Tomorrow. Archway Publishing. 2016. 186 p.
2. Budzko V.I., Leonov D.V., Nikolaev V.S., Onykijj B.N., Sokolina K.A. Razvitie informacionno-analiticheskojj podderzhki nauchno-tekhnicheskojj dejatelnosti v Nacionalnom issledovatelskom jadernom universitete «MIFI» // Sistemy vysokojj dostupnosti. 2011. T. 7. № 4. S. 4−17.
3. Sokolov I.A., Budzko V.I., Kalinichenko L.A., Sinicin I.N., Stupnikov S.A. Razvitie rabot v oblasti «Bolshikh dannykh» v Rossijjskojj akademii nauk // Sistemy kompjuternojj matematiki i ikh prilozhenija. 2015. № 16. S. 103−110.
4. SHouman M.A. Mnogojazykovyjj informacionnyjj poisk s ispolzovaniem multiagentnojj platformy. Dis. - kand. tekhn. nauk. M.: 2015. 132 s.
5. Onykijj B.N. i dr. Nauchno-tekhnicheskijj otchet o vypolnenii 5 ehtapa Gosudarstvennogo kontrakta № 16.740.11.0129 ot 2 sentjabrja 2010 g. i Dopolnenie ot 29 fevralja 2012 g. № 1. 2012.
6. Beljaev K.V., Bosov A.V., Krajushkin D.V. Obzor i sravnitelnyjj analiz informacionno-analiticheskikh sistem. M.: IPI RAN. 2008. 136 s.
7. Vassil Sgurev, Ronald R. Yager, Janusz Kacprzyk, Vladimir Jotsov Innovative Issues in Intelligent Systems. Springer. 2016. 353 p.
8. Artamonov A.A., Nikolaev V.S., Pronicheva L.V., Sokolina K.A., Ushmarov I.A. Vizualizacija semanticheskikh otnoshenijj v multiagentnykh sistemakh // Nauchnaja vizualizacija. 2014. T. 6. № 3. S. 68−76.
9. Charu C. Aggarwal Data Mining: The Textbook. Springer International Publishing. 2015. 734 p.
10. Artamonov A.A., Galin I.JU., Leonov D.V., Mikhina E.K., Onykijj B.N., Sokolina K.A. Poiskovye agentnye tekhnologii s mnogojazychnym tezaurusom // Vestnik NIJAU «MIFI». 2015. T. 4. № 4. S. 369−376.