TREC Blog and TREC Chem: A View from the Corn Fields

This year the Blog Track contained two tasks: Top Stories Identification and Faceted Blog Distillation Tasks. Our submissions for both tasks are described below. In this, our first entry into the blog track, we explore various strategies (latent Dirichlet relevance model, URL based ranking, query expansion etc.) for both tasks. We first indexed the blog data with Lucene and identified occurrences of Headline URLs in the permalink documents (which included the content of the posts as well as the side bars of the web pages). Text windows (+/800 characters including HTML code) surrounding the occurrences were harvested. The four runs submitted for the first task and the two for the second are described below.