Modified collaborative filtering for hybrid recommender systems and personalized search: The case of digital library

Abstract

Digital libraries constitute a considerable source of digital content providers, similar to video and music streaming services. Therefore, a solid, reliable and intelligent recommender system is essential to accommodate the plethora of different interests amongst its users. In view of this compelling need, this paper presents a modification to the classic collaborative filtering technique which incorporates the user’s actions into the recommendation production process. In this way, the user implicitly provides extra data to the collaborative filtering-based recommender system, resulting in higher quality recommendations and personalized search results, especially when combined with elements of content-based filtering. The results of the above-mentioned modification are presented by integrating the recommender system to a web-based digital lending library application. The evaluation of the application was made using the inspection method of cognitive walkthrough.