Nowadays, Artificial Intelligence (AI) is applied to many fields and enhances the quality of human life. For the movie industry, there is the need for smarter systems that can understand user’s behaviours and give more appropriate responses. Along with this goal, the thesis “AI Study for Building Chatbot and Movie Recommendation System” was coined. The objective of this topic is to construct a website for quickly looking up movie information according to the user's preference, as well as enable human-machine communication via chatbox.
The front end of the website is built with React library. Meanwhile, its backend is accomplished by using NodeJS, which is responsible for processing and access to the NoSQL MongoDB database.
The user intents classification is the most important task of chatbot because it affects the chatbot ‘s response behavior. The Multi-class SVM algorithm is used for model training and user intents classification thanks to its higher accuracy compared to other algorithms.
To suggest movies, the thesis used a method combining both Content-based Filtering and Collaborative Filtering that comes up with a list of movies that are best suited to each user, based on their rating history or their watched movies list.
The result of the thesis is a website with two privilege types, one is for users and the other is for administrators. Users can search movies information, seek, rate movies, send messages to support chatbot and receive movie recommendations. Administrators can manage data information such as movies, genres, message, etc. Moreover, they can train model chatbot to improve chatbot ‘s knowledge. Recommendation system and chatbot are also successfully integrated into the website.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent