Article Preview
Top1. Introduction
The advent of WWW and concomitant increase in online information have caused information overload and ignited research in Recommender System (RS) (Perugini et. al., 2004). RSs are a subclass of information filtering system that are primarily directed towards individuals who lack sufficient personal experience or competence to evaluate the potentially overwhelming number of alternative items offered by a system (Francesco et. al., 2011 and Resnick, P., & Varian, 1997). RSs have become extremely common in recent years, and have been applied in a variety of applications.
The large amount of product information available online today poses vast challenges to both customers and online businesses in sighting products that best fits their needs. Obtaining recommendations from trusted sources is a critical component of natural process of human decision making. Hence, RSs are utilized to generate meaningful recommendations to a collection of users for items or products that might interest them so as to minimize the time spent while searching (Melville & Sindhwani, 2005).
As information and e-commerce burgeons, buyers and sellers are faced with a number of challenges including fuzziness in choice making and the challenge of personalizing advertisement efforts (Omisore et al., 2013). RS applies data analysis techniques to support users in identifying items of interest among large item-sets (Ghauth & Abdullah, 2010). RSs typically produce a list of recommendations using one or more recommendation algorithms. Such algorithms use different types of data to create possibility for subtle personalization in which completely organic personalized experience is presented to users.
Among popular RS techniques, Content Based Filtering (CBF) and Collaborative Filtering (CF) are commonly used to recommend products, meticulously, to users. CF is the most widely used method to recommend items for users (Breese et. al., 1998). It makes recommendation according to users’ similarity base on items which are rated by the active user (Haifeng et. al., 2014). Memory based method is a CF method that gives considerable accuracy but its computing time grows rapidly with the increasing number of users and items hence, it is difficult to respond in some real-time conditions. Model based method, which is an alternative CF technique tends to be faster in prediction time but its performance is not as good as the former. CBF techniques are used to match the attributes of items (of interest) with the characteristics of an active user. The technique makes use of discrete characteristics inherent to the item so as to recommend additional items with similar attributes (Mooney & Roy, 2000). A major limitation of CBF is its inability to learn from a user's preferences from actions of a certain content and adapt it for some other users.
Various RS techniques had been explained in (Resnick & Varian, 1997; Sarwar et al., 2001; Melville et al., 2002; and Ojokoh et al., 2012). Another popular RS technique is Knowledge Based Filter (KBF) which can be used to predict products by making inferences from users’ needs and preference. KBF has capability to minimize the portfolio effects inherent to the former techniques however, it suffers a major drawback found in other knowledge-based systems. Hybrid RSs have been developed by combining two or more techniques to provide more suitable system that has better performance with fewer drawbacks. Such systems have been emphasized in (Resnick & Varian, 1997 and Omisore & Samuel, 2014).
RSs for research papers help researchers keep track of their research field despite the rate of growth in scientific research (Vellino & Zeber, 2007). Therefore, this paper proposes a hybrid recommendation approach for personalized retrieval of research articles. The approach integrates Content Based, Collaborative, and Knowledge Based filtering techniques to improve researchers’ selection accuracy while surfing Internet for research articles to be purchased. The proposed RS is to help researchers reduce the time taken to search for useful articles and therefore minimizes blind purchasing acts.