An extracting model for constructing actions with improved part-of-speech tagging from social networking texts

The recent viral growth of social network systems such as Twitter, Facebook and MySpace have created many interesting and challenging problems to the research community, which enable to perform context aware-reasoning. Social networking is a set of social actors (individuals or organizations) that are connected to provide a set of interaction. We consider, in this paper, the problem of information extraction from social networking specially Twitter and Facebook. To extract text from social networking, we need several lexical features and large scale word clustering. We attempt to expand existing tokenizer and to develop our own tagger in order to support the incorrect words currently in existence in Facebook and Twitter. Our goal in this work is to benefit of the lexical features developed for Twitter and online conversational text in previous works, and to develop an extraction model for constructing a huge knowledge based on actions.