| Summary: | 碩士 === 國立交通大學 === 電信工程研究所 === 103 === With the prosperity of e-commerce, online vendors use the recommendation systems in different fields. Classic algorithms for data analysis, such as cosine-similarity, user-based collaborative filtering, are designed assuming that data are stationary and will not change over time. Since the scale and variability of data growing gradually, these methods will encounter the issues of the memory deficient and the out-of-date model, which degrade the recommendation accuracy intensively. In addition, retraining the whole model for every new arrival record results in high complexity. We propose a light-weight adaptive updating method to overcome these issues. Comparing with the explicit feedback recommendation asking the customers to express their opinions on the recommended items, the implicit feedback recommendation is easier to collect and non-intrusive source. However, the dynamic time-variant system with implicit feedback has not been seen in the literature.
In this theory, we propose a stochastic gradient descent based real-time incremental updating method (RI-SGD) for matrix factorization to deal with a time-variant system based on the implicit feedback. We compare our method with retraining the whole model and show that our method costs less than 1\% of the retraining time with a competitive accuracy.
|
|---|
