DUTIR at TREC 2007 Blog Track

This paper describes DUTIR at TREC 2007 Blog Track. In data preprocessing, a non English language list created from the corpus was used to remove the non English blogs, blog templates were also used to extract the post and comment; in Opinion Retrieval task, information in the meta tags were also indexed; in the polarity subtask, a method based on SVM was used and the Information Gain attribute selecting method was used to assist SVM; in Feed Distillation task, three type of feeds were analyzed according to their tag structure, information extracted from particular tags of the feeds were finally indexed.

• Type: Conference
• Year: 2007
• Where: TREC
• Authors: Rui Song, Qin Tang, Daming Shi 0002, Hongfei Lin, Zhihao Yang
• Keywords: Attribute Selecting Method, Opinion Retrieval Task, TREC 2007, TREC 2007 Blog, TREC 2008