Performance Of The Recommendation System Using Temporal Event Data

Abstract

Although RS has issues including idea drifts and temporal dynamics, it has become popular since it helps customers by suggesting things they might like based on their past purchases and preferences. Due to problems with concept drift, traditional RS approaches do a poor job of providing precise concepts, but they are great at generating recommendations. In light of these issues, a great deal of research has been carried out. however, you'll have to put in a lot of time and effort to fix the interest drift. It is not possible to track user preferences in real-time. This approach depicts the situation in a static and imperfect light. We suggest that in order to make the efficacy of recommender systems more understandable, researchers should calculate metrics over longer time periods (e.g., weeks or months) and display the results graphically (e.g., with a line chart). Insightful predictions regarding the future performance of an algorithm can be derived from results demonstrating how its efficacy changes over time. It is possible to make better informed decisions on the algorithms to utilize in a recommender system if we collect more data on their performance over time, identify trends, and make more precise predictions about their future performance.