Implementing Background Net with Knowware System for Personalized Keyword Support

Abstract

This article proposes a novel approach that combines user’s instant requirement described in keywords with her or his long-term knowledge background to better serve article selection based on personal preference. The knowledge background is represented as a weighted undirected graph called background net that captures the contextual association of words that appear in the articles recommended by the user through incremental learning. With a background net of user constructed, a keyword from the user is personalized to a fuzzy set that represents contextual association of the given keyword to other words involved in the user’s background net. An article evaluation with personal preference can be achieved by evaluating similarity between personalized keyword set based on user’s background net and a candidate article. The proposed approach makes it possible to construct a search engine optimizer running on the top of search engines to adjust search results, and offer the potential to be integrated with existing search engine techniques to achieve better performance. The target system of personalized article selection can be automatically constructed using Knowware System which is a development tool of KBS for convenient modeling and component reuse.