Structuring T weets for improving T witter search

Spam and wildly varying documents make searching in T witter challenging. Most T witter search systems generally treat a T weet as a plain text when modeling relevance. However, a series of conventions allows users to T weet in structural ways using a combination of different blocks of texts. These blocks include plain texts, hashtags, links, mentions, etc. Each block encodes a variety of communicative intent and the sequence of these blocks captures changing discourse. Previous work shows that exploiting the structural information can improve the structured documents (e.g., web pages) retrieval. In this study we utilize the structure of T weets, induced by these blocks, for T witter retrieval and T witter opinion retrieval. For T witter retrieval, a set of features, derived from the blocks of text and their combinations, is used into a learning‐to‐rank scenario. We show that structuring T weets can achieve state‐of‐the‐art performance. Our approach does not rely on social media features, but when we do add this additional information, performance improves significantly. For T witter opinion retrieval, we explore the question of whether structural information derived from the body of T weets and opinionatedness ratings of T weets can improve performance. Experimental results show that retrieval using a novel unsupervised opinionatedness feature based on structuring T weets achieves comparable performance with a supervised method using manually tagged T weets. Topic‐related specific structured T weet sets are shown to help with query‐dependent opinion retrieval.