The Potential Knowledge Recommendation System using User’s Search Logs

Abstract

Background/Objectives: This paper proposes a potential query recommendation system based on the user search history so that information search system users can express their potential information needs in a query, and the information they want can be searched. Methods/Statistical Analysis: The proposed system used users’ search query to analyze the associative relationship with existing users’ search history, and extracted users’ potential information needs. The extracted potential information needs are recommended to users in the recommendation query. Findings: This paper used 27,656 pieces of search history data for analyzing the utility of the proposed system and conducted a behavioral experiment. The experiment found that the subjects showed a statistically higher level of satisfaction when using the proposed system than when using a general search engine. Improvements/Applications: In the future, it will be possible to secure the reliability of recommended queries by expanding and solidifying the search history through researches on personalization.