Using temporal evidence in blog search

• Autor(a) principal: Sérgio Nunes
• Data de Publicação: 2009
• Outros Autores: Cristina Ribeiro, Gabriel David
• Tipo de documento: Livro
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/26928
• topic: Tecnologia da informação, Engenharia de computadores, Ciências da computação e da informação; Information technology, Computer engineering, Computer and information sciences

In this paper we present a study on the relevance of web documents over time and the use of temporal evidence in blog search tasks. Time is an intrinsic property of social media, most notably in blogs where each post is typically attached with a timestamp representing its publish date. However, due to the challenges in obtaining document collections containing temporal information, research on this field has been scarce. We base our study on the Blog06 collection and the relevance assessments produced in the context of the TREC Blog Track, to investigate the relevance of time-based features in standard retrieval tasks. We observe small, but statistically significant improvements over a BM25 baseline when temporal information is used. Also, we find a direct connection between recency and relevance of documents for ad-hoc retrieval.