SyntcRec: a Syntactic Recommender System Based on Improved Feature Selection Technique in Large Scholarly Data

Abstract

In recent years, the amount of academic literature has been increasing, which results in information overloading. Another result of this fast data progress is that, one can feel lost in the variety of literature and realize the difficulties to yield decisions. To solve these issues, it is needed to filter, rank and competently deliver relevant material to several Internet operators. Recommendation schemes rectify this problem by providing users with personalized content and services. This paper presents a novel recommender system called syntactic Recommender, which employs improved feature selection techniques, a combination of ACO-GA (Ant Colony Optimization-Genetic Algorithm) algorithms aimed at feature selection in order to raise the accuracy of recommendations in recommender systems. The paper focuses on two key concepts, the first is performing and giving more exact object predictions to the operator and the second is handling a huge volume of data. The aim of this paper is to offer recommendation outcomes based on the users requests with accuracy and competency. The proposed ACO-GA feature selection algorithm and the Cosine Similarity technique increase accuracy of the Recommender system up to 92%.