A Generative Blog Post Retrieval Model that Uses Query Expansion based on External Collections

摘要: User generated content is characterized by short, noisy documents, with many spelling errors and unexpected language usage. To bridge the vocabulary gap between user's information need documents in a specific user environment, blogosphere, we apply form of query expansion, i.e., adding reweighing terms. Since blogosphere noisy, expansion on collection itself rarely effective but external, edited collections are more suitable. We propose generative model for expanding queries using external which dependencies queries, explicitly modeled. Different instantiations our discussed make different (in)dependence assumptions. Results two (news Wikipedia) show that retrieval effective; besides, conditioning very beneficial, making candidate terms dependent just document seems sufficient.