| This dissertation covers two related topics---personalized search and business relationships discovery---in the area of knowledge discovery from the Web.; In part I, we propose an automatic personalized search approach that categorizes search results under a user's interests. In two sets of controlled experiments, we compare our personalized categorization system with a list interface system and a nonpersonalized categorization system, respectively, on the basis of the type of task and query length, and identify conditions under which our proposed system outperforms each of the baseline systems.; In part II, we present a news-driven, social network analysis based business relationship discovery framework and study two different business relationships, company revenue relation and competitor relationship. We find that the structural attributes derived from the intercompany network are valuable in predicting the business relationships. Also, our proposed framework is scalable (as compared to manual approaches) and language-neutral. |
