A novel recommender system for E-commerce
Pang-Ming Chu, Shie-Jue Lee · 2017
Recommender systems play an important role in human lives nowadays. They have been used in many electronic commercial activities, and are growing more popularly due to the development of IOT (Internet of Things), big data analysis, and machine learning techniques. However, most recommender systems are only based on the user-item rating matrices which are usually very sparse. The lack of information leads to bad recommendation results. Furthermore, a large number of users and items are usually involved in the recommendation process. These extremely high dimensional data may thwart the efficiency of a recommender system. In this paper, we propose a novel method to overcome these problems. Word2Vec is adopted to extract information from the comments users have made on the items bought. Then dimensionality reduction is applied to project the acquired data into a lower dimension space. A clustering algorithm is then used to group the involved items to a small number of clusters. Finally, the recommendation results are generated for each user. The effectiveness of our proposed method is demonstrated by the results of experiments with some real world data sets.