Query Expansion Based on the Distance Constraint Activation of Human Memory

Abstract

Query Expansion is usually used to enhance user's query in document retrieval and knowledge recommender system. It aims to help user extend the original query to resolve the problems of user's fuzzy demand and limited domain knowledge when user faces complex knowledge demand. Current researches extend user's query mainly based on statistical method while the characteristics of human memory retrieval process are seldom taken into account. However, the process of human memory retrieval has good ability of extending the query when human faces fuzzy demand and limited domain knowledge. In this paper, we put forward a new model for Query Expansion based on the Distance Constraint Activation model of human memory. We build up the ALN (Association Link Network) of documents corpus to imitate human memory network and do the Distance Constraint Activation process on it in order to offer user Query Expansion service in accordance with the pattern of human memory retrieval process. Comparing with the traditional Query Expansion methods, our method is able to provide the semantic relation expansion and the semantic community expansion to the user. All these expansions aim to help user understand his/her demand or extend his/her knowledge in the domain corpus more easily. Several examples given in the experiments show that the semantic-close terms our method provides make the expansion terms more semantic coherent. And the semantic relations and semantic communities can give user a brief description of the knowledge to further understand his/her demand and extend his/her domain knowledge.