CONTENT-BASED RECOMMENDATION USING MACHINE LEARNING

Currently,the user profile based online recommender system has become a hit both in research and engineering domain. Accurately capturing users' profile is the key of recommendation. Recently, lots of researches on user profile extraction have been launched, including content-based recommendation. To better capture users' profiles, a three-step profiling method is adopted in this work. (1) Purchase item prediction is made based on Logistic Regression. (2) Purchase category prediction is made based on support vector machine (SVM), and (3) User's rating prediction is made based on convolutional neural network (CNN) and Long Short-Term Memory (LSTM). This work outperformed the baseline model on the user dataset collected from Amazon. So, in conclusion, the work has the ability of giving reasonable recommendation for users who would like to purchase online. In the future, the video signal processing techniques will also be taken under consideration to capture users' face expression for better recommendation.

Automated summary

The study introduces a user profile-based recommender system with a three-step profiling method, including purchase item prediction, purchase category prediction, and user rating prediction. Results show that this method outperformed the baseline model on user dataset, providing reasonable recommendations for users. Future considerations include the use of video signal processing techniques to capture users' facial expressions.