Expertise analysis of information senders of Web pages

In this study, we address the problem of searching experts in an arbitrary topic on the Web. In particular, we propose two methods that analyze the expertise of information senders of Web pages: 1) a method that computes expertise score based on hit count from a search engine (hit count method), and 2) a method that computes expertise score based on the number of documents that are attributed to an information sender (attribution count method). We evaluated both methods using a crawl and search infrastructure which indexes 120 million Japanese Web pages. The results show that the attribution count method outperforms hit count method in terms of precision and processing time.