An Innovative Approach: Typicality-Based Collaborative Filtering for Recommender Systems
Veera Talukdar, Vivek Veeraiah, Devika Rani Roy, Jay Kumar Pandey, Ankur Gupta, Dharmesh Dhabliya
Abstract
There are two types of collaborative filtering (CF) methods: user-based and item-based. CF is a frequently used technique for recommender systems. Item-based CF concentrates on making suggestions based on strong correlations, whereas user-based CF concentrates on discovering comparable user preferences and doing so. Today, the majority of CF approaches employ corated objects to determine how similar users are. This study suggests a typicality-based method called TyCo that identifies users who are “neighbors” of one another by looking at user group typicality. TyCo represents a user through a user typicality vector, which identifies their preferences for various goods. In order to get around the drawbacks of conventional collaborative filtering techniques, it chooses “neighbors” by comparing users' similarity based on their typicality levels. This is the first piece of work to use typicality to collaborative filtering, and it runs experiments to assess and show off its benefits. The cost of preprocessing steps, which include creating user prototypes and assessing user typicality in user groups, varies based on the clustering approach selected.