Knowledge extracting from micro-blog topic

Abstract

Micro-blog topic knowledge extraction is the process of extracting knowledge hiding in sparse micro-blog data published by users. This paper presents a microblog topic knowledge extraction method. Firstly, the micro-blog text is preprocessed by removing the non-meaningful text (e.g. web links and animations). Secondly, part-of-speech weight and First and Last Word Weights (FLWW) are used to extract keywords candidate set which can become some entities of main knowledge (e.g., diaper and beer in diaper → beer) of a micro-blog. Finally, the non-entity keywords which are not main meanings of a micro-blog in keywords candidate set are filtered by the left and right entropy, and then the knowledge is extracted by the association rules between keywords. Experiments show that the algorithm is superior to the traditional algorithm.