Ranking semantic relationships between two entities using personalization in context specification

The experimental results prove that the proposed method is more efficient for ranking semantic relationship paths, and the effectiveness of the ranking method is demonstrated through Spearman's foot rule correlation and precision rate.

Abstract

Semantic relationship deals with complex relationship between two entities in a knowledge base. In the process of finding the relationship, multiple paths connecting entities could be explored. Each path has a different meaning depending on the type of relation. Some of them may be relevant while others may be irrelevant depending on users' perspective with varying results for the same query. Hence, ranking these paths according to the user's need is imperative. In this paper, we adopt personalization approach to rank the semantic relationship paths. Here, user's interest level in various domains can be captured through their web browsing history and it is incorporated in calculating the context weights of the paths during the ranking process. The effectiveness of the ranking method is demonstrated through Spearman's foot rule correlation and precision rate. The experimental results prove that our proposed method is more efficient for ranking semantic relationship paths.