Finding keywords in blogs: Efficient keyword extraction in blog mining via user behaviors

Readers are becoming accustomed to obtaining useful and reliable information from bloggers. To make access to the vastly increasing resource of blogs more effective, clustering is useful. Results of the literature review suggest that using linking information, keywords, or tags/categories to calculate similarity is critical for clustering. Keywords are commonly retrieved from the full text, which can be a time-consuming task if multiple articles must be processed. For tags/categories, there is also a problem of ambiguity; that is, different bloggers may define tags/categories of identical content differently. Keywords are important not only to reflect the theme of an article through blog readers’ perspectives but also to accurately match users’ intentions. In this paper, a tracing code is embedded in Blog Connect, a newly developed platform, to collect the keywords queried by readers and then select candidate keywords as co-keywords. The experiments show positive data to confirm that co-keywords can act as a quick path to an article. In addition, co-keyword generation can reduce the complexity and redundancy of full-text keyword retrieval procedures and satisfy blog readers’ intentions.