Using text mining algorithms in identifying emerging trends for recommender systems

Recommendation systems as the main e-commerce tools play an important role in business survival. Therefore, recommender systems and their challenges are a concern for scholars and professionals. Since this kind of system offers appropriate suggestions to online users using their interests and preferences, a lack of information about users and their purchase histories has negative impacts on the performance of recommender systems. This issue is known as “cold start problem” including cold-start user as well as cold start item and occurs when a new user logs in or an item is registered newly in a system. To deal with this problem, a lot of scientists have started studying and have done great researches annually. The first and most important step to optimize recommender systems is to have enough knowledge about previous studies and their proposed methods and algorithms using a review of these researches. Collecting and reading each of these articles is a difficult and time-consuming process. Accordingly, in this paper, we analyze the textual data collected from the best journal articles addressing the challenges of recommender systems to identify new and emerging fields in this area. This research can pave the way for future researchers of this field to develop more and more recommendation systems. The way to conduct this research is to first extract valid scientific articles in the domain of recommender systems challenges from the reputable scientific databases, the web of science. Then, using different text mining algorithms on keywords, titles, and abstracts of these articles, identification of emerging topics in this field is achieved.