KLE at TREC 2008 Blog Track: Blog Post and Feed Retrieval
Yeha Lee, Seung‐Hoon Na, Jun-Gi Kim, Sang-Hyob Nam, Hun-Young Jung, Jong-Hyeok Lee · 2008
This paper describes our participation in the TREC 2008 Blog Track. In the opinion task, we made the opinion retrieval model that consists of the preprocessing, the topic retrieval, the opinion finding, and the sentiment classification part. In topic retrieval, our system is based on the passage-based retrieval model and feedback. For the opinion analysis, we create a pseudo opinionated word (POW), O, which is a representative of all opinion words, and expanded the original query with O. Our approach for the blog distillation is based on two resource selection techniques. We integrated the average score of all posts within the feed, and the average score of the high top N post scores. We additionally examine the pseudo-relevance feedback for distillation task by focusing on various document selection schemes to expand query terms. The experimental results show the significant improvement over previous results. 1