Methods for ranking user-generated text streams

摘要: User generated content are one of the main sources information on Web nowadays. With huge amount this type data being everyday, having an efficient and effective retrieval system is essential. The goal such a to enable users search through retrieve documents relevant their needs. Among different tasks user content, retrieving ranking streams important ones that has verious applications. task rank streams, as collections with chronological order, in response query. This than traditional where single temporal properties less ranking. In thesis we investigate problem user-generated case study blog feed retrieval. Blogs, like all other have specific require new considerations methods. Blog can be defined blogs recurrent interest topic given We define three each which introduces challenges task. These include: 1) term mismatch retrieval, 2) evolution topics 3) diversity posts. For these properties, its corresponding propose solutions overcome those challenges. further analyze effect our performance system. show taking into account for developing help us improve state art proposed methods, specifically pay attention believe any streams. when combined content-based information, useful situations. Although apply methods they mostly general applicable similar stream problems experts or twitter users.