Article Preview
TopIntroduction
Everyday huge volumes of information are getting accumulated to cater the needs of growing web users. In this scenario, for a common man traversing through this enormous pile of information and getting the needed information continues to be a complex task. In order to simplify the task of searching the needed information from the web, researchers developed recommender system. Generally, most of the recommender system uses collaborative filtering algorithm to predict recommendations for a user (Su & Khoshgoftaar, 2009) (Konstan & Riedl, 2003) (Linden, Smith, & York, 2003) (Harper & Konstan, 2016) (Koschmider, Hornung, & Oberweis, 2011). Collaborative filtering (CF) algorithm utilizes the user information to find the neighbour with highest similarity (Herlocker, Joseph A, & Riedl, 2000) (Herlocker, Konstan, Borchers, & Riedl, 1999). The item ratings given by the neighbours are computed to render list of recommendation to the user (Resnick, Iacovou, Suchak, Bergstrom, & Riedl, 1994). However, when a new user enters the system CF algorithm fails to process recommendations due to lack of information about the user resulting in new user cold start problem (Victor, Cornelis, Teredesai, & Cock, 2008) (Son, 2016) (Khusro, Ali, & Ullah, 2016) (Pereira & Varma, 2019). New user cold start problem is defined as the inability of the system to render recommendations due to the unavailability of information about the user (Shi, Hu, Zhao, & Philip, 2019). Even though, researchers came up with solutions to alleviate the persisting new user cold start problem, yet there is a huge room for improvement (Son, 2016) (Chen, Wan, Chung, & Sun, 2013) (Moses & Babu, 2018).
The existing solutions tries to fetch the user related information from other third-party sources or tries to cluster the user to specific group based on minimal user rating information. Other methodologies like asking a newly visited user to fill some survey forms, or give ratings and even asking them to authenticate other social web platforms to get their information will annoy the user at certain point. Also it becomes recommendation systems responsibility to gain user trust and assist user so if user starts to believe the system is wrong then it will be a total failure of the recommendation system primary objective. Therefore, in this article, a genetic algorithm-influenced recommendation system acting on user gender and movie genre information is proposed to alleviate the cold start problem specifically in movie recommendation system.
By employing the genetic evolution principles on genre and user gender, movies that interest the user is sorted out from the huge amount of information. After computing, when a new user enters the system based on the user gender, interesting as well as unique item will be recommended to the user. Genetic algorithm is used widely in searching the best solution among the various possible solutions to a certain problem (Goldberg, 1989) (Michalewicz, 2013) (Ribeiro Filho, Treleaven, & Cesare, 1994). The effectiveness of genetic algorithm in finding the optimal solution made researchers to adapt genetic algorithm (GA) influenced procedures to solve the optimization problems belonging to various domains (Maulik & Bandyopadhyay, 2000) (Leu, Yang, & Huang, 2000) (Eberhart & Shi, 1998).