New Histogram-Based User and Item Profiles for Recommendation Systems

A recommendation system is used as a tool for data filtering that works on the principle of discovering patterns in data on customer behaviour to suggest the most relevant items to a user. Based on the existing trend- content based, collaborative filtering, hybrid model are the most popular techniques that are used for recommendation systems. By using Symbolic Data Analysis (SDA) [9], we extended the approaches of the content-based and collaborative filtering techniques in our work. Standard data analysis has been enhanced by SDA, that takes symbolic tables as input and results in symbolic objects as output. The symbolic data is highly used to aggregate and reduce the loss of data. The similarity scores will be generated for each technique and the NDCG metrics is used to evaluate the performance of these models. Finally, we have observed that SDA is significantly helpful for understanding and modelling of the histogram data in optimizing the performance and generating the recommendations accurately.