The following article considers the problem of web-surfing automation and content filtration. The principal objective of this project is to develop a software solution to this problem - a multi-agent system for analyzing VKontakte users’ interests - and providing a recommendation system EZSurf. The article describes the development and application of a multi-agent recommender system EZSurf that performs analysis of interests and provides recommendations for the social network VKontakte users based on the data from the profile of a particular user. The article also provides an analysis of different methods, technological solutions, and similar products aimed at content filtration, as well as their advantages and disadvantages. EZSurf allows automating the web-surfing process and content filtration with the use of user’s profile in a particular social network to collect data and API of external services (LastFM, TheMovieDB). For search and selection of information an agent (Recommender) that works as web-crawler has been implemented. Such an approach contributes to optimization of the recommender system, because it does not require creation of its own object classification system and objects database. The functionality of multi-agent system was separated between three agents. The first agent (Collector) collects user data from “VKontakte” profile using VK API. The second agent (Analyzer) collects similar objects from databases of external services (LastFM, TheMovieDB) that will be the criteria for further search of recommendatory content. The third agent (Recommender) based on the principle of a search robot is used for searching content. System «EZSurf» can be exploited by the users of social network “VKontakte” in everyday life to save time on web-surfing. At the same time the users will get recommendations on the content filtered depending on preferences of every particular user. The system can be further developed. There are several ways for its evolution: extension of sites registry, usage of other data from profile in addition, optimization of the algorithm for web-pages indexing and content parsing (Recommender), data collection from the several social networks.
mul'tiagentnaya, rekomendatel'naya sistema, sistema, fil'traciya soderzhimogo, social'nye seti, rekomendatel'nyy kontent, veb-serfing
1. Glibovec N.N. Ispol'zovanie JADE (Java Agent Development Environment) dlya razrabotki komp'yuternyh sistem podderzhki distancionnogo obucheniya agentnogo tipa. URL: http://www.ebiblioteka.lt/resursai/ Uzsienio%20leidiniai/IEEE/Russian/2005/Nr%203/OTO_2005_3_04.pdf
2. Zhuravskiy V. Prostoy poiskovyy robot. URL: http://juravskiy.ru/?p=1005
3. Ohapkina E.P., Lukoyanov I.A., Voronov V.I., Voronova L.I. Razrabotka i vnedrenie poiskovogo robota dlya analiza interesov klientov // Studencheskiy nauchnyy forum: VI Mezhdunarodnaya studencheskaya elektronnaya nauchnaya konferenciya (15 fevralya - 31 marta 2014 goda). URL: http://www.scienceforum.ru/2014/ 495/4758
4. Pestryaev A.A., Voronova L.I. Mul'tiagentnaya sistema. Vzaimodeystvie agenta-sobiratelya s bazoy dannyh // Studencheskiy nauchnyy forum: VI Mezhdunarodnaya studencheskaya elektronnaya nauchnaya konferenciya (Elektronnyy resurs), 15 fevralya - 31 marta 2014 goda. URL: http://www.scienceforum.ru/2014/495/4036
5. Ekstrand M.D., Riedl J.T., Konstan J.A. Collaborative Filtering Recommender Systems // Retrieved January 26, 2014, from Social Computing Research at the University of Minnesota website. URL: http://files.grouplens.org/ papers/FnT%20CF%20Recsys%20Survey.pdf
6. Lane J. Musicians Find Fans At Grooveshark Artists. URL: http://blogcritics.org/musicians-find-fans-at-grooveshark-artists/
7. Ricci F., Rokach L., Shapira B. Introduction to Recommender Systems Handbook, Recommender Systems Handbook, Springer.
8. Sahebi S., Cohen W.W. Community-Based Recommendations: a Solution to the Cold Start Problem. URL: http://d-scholarship.pitt.edu/13328
9. Surfingbird. Oficial'nyy sayt kompanii. URL: http://surfingbird.ru/about REFERENCES
