Practical approach for recommender systems
Prathamesh S Tugaonkar, Vidya Chitre · 2016
Every day, number of pages gets added on web which makes tracking of their links cumbersome. Due to this, problem of overloaded data has come up. This issue led researchers to thoroughly go through different aspects of Web Usage Mining (WUM). Another issue of traditional system is of recommendations which are also a part of WUM and Web logs. This paper proposed a system of recommendations which uses tokenization to separate the users and information is conveyed to Resource Description Framework (RDF) for Semantic data generation and display. Session based clusters are formed and frequency of items is noted. Algorithms are also proposed for smooth working of the system. Web log information is needed for understanding the general mentality or behaviour of the user. We have also proposed Natural Language Processing (NLP) techniques for conditions where users preferences regarding products are not generated.