Current Proceedings on Technology

Yazarlar: Mohammad Reza Fani Sani, Amin Omidvar, Farzad Farahmandnia, Saeed Shiry

Konular:-

Anahtar Kelimeler:User Clustering,Social Networks,WordNet,SOM,Text Mining, 

Özet:   Nowadays, social networks are the most successful applications on the Internet. Numerous people with different characteristics in terms of goal, culture, education and nationality are the members of social networks and do wide range of activities on them such as content sharing, communication and entertainment. Since huge volume of data about social networks' members exists on them, there is a good opportunity for market segmentation, efficient advertising and customization through employing text mining methods in order to categorize members based on their interests. In this paper a novel method which is called WSOM is introduced in order to cluster social networks' members based on their sent posts. To do that, first TFIDF was calculated for all posts of each member. Then by utilizing information gain, best features were selected. After that by using SOM members were clustered in different groups. Topic of each cluster was defined by employing WordNet ontology. In order to verify the proposed method, WSOM was applied on real Friendfeed's data which is the one of the most famous social networks on the Internet

---