Microblog summarization

摘要

Various embodiments provide summarization techniques that can be applied to blogs or microblogs to present information that is determined to be useful, in a shortened form. In one or more embodiments, a procedure is utilized to automatically acquire a set of concepts from various sources, such as free text. These acquired concepts are then used to guide a clustering process. Clusters are ranked and then summarized by incorporating sentiment and the frequency of words.

主权项

1. A method comprising:
processing, using a computing device, multiple resources to build a word dictionary configured to enable summarizing a plurality of microblogs, the word dictionary containing individual words that are nouns that are associated with a particular domain; using, using the computing device, the word dictionary to create concepts, at least some individual concepts comprising a semantic tag comprising multiple words; assigning, using the computing device, a plurality of microblogs to a plurality of the concepts effective to form potential clusters; computing, using the computing device, a membership score for each microblog/cluster pairing; using, using the computing device, the membership score to assign a microblog to a cluster; ranking, using the computing device, each cluster using an entropy measure that incorporates sentiment value computed for each microblog and probability of words computed over microblogs in a particular cluster, and summarizing the plurality of microblogs by displaying a cluster summary for each cluster on a display of the computing device.