Strategies to Map Play Time in Steam Platform Dataset into 5-Star Ratings

Abstract

This paper presents strategies to map play time in Steam dataset into 5-Star rating. Our objective is to check whether we could use the created explicit feedback in recommender system. In this study we calculated z-score for each users' duration of playtime and using the score we applied 3 strategies to map duration of playtime into explicit feedback. These strategies are Same Size Bin (SSB), Breakpoint Based Bin Sized (BBS) and Hybrid Based Bin Size (HBS). According to RSME value and Hit Rate, we found that the best strategies to map duration of playtime into explicit feedback is using HBS strategies. We checked the feasibility of using the created explicit feedback by feeding it to KNN algorithm. We compare RSME from others researches that used the same algorithm and found the RSME value that we get are in the same range as the researches. Based on this observation, it can be concluded that it is feasible to use created explicit feedback in Recommender System.