Recommender System for Multiple Databases Based on Web Log Mining

Abstract

Finding information from a large collection of resources is a tedious and time-consuming process. Due to information overload, searchers often need help and assistance to search and find the information. Recommender system is one of the innovative solutions to the problem related to information searching and retrieval. It helps and assist searchers by recommending the possible solution based on the previous search activities. These activities can be obtained from the web log, which requires a web log mining approach to extract all the keywords. In this study, keywords obtained from the library web log were analysed and the search keyword patterns were obtained. These keyword patterns were from several databases or resources that were subscribed by the library. The finding revealed some of the popular keywords and the most searchable databases among the searchers. This information was used to design and develop the recommender system that can be used to assist other searchers. The usability test of the recommender system showed that it is beneficial and useful to the searchers. These findings will also benefit the management in planning and managing the subscription of online databases at the university’s library.