Web application movie recommendation (optimize interaction and data analyzation on wed data)

Abstract

In technology era 4.0, the entertainment on the internet has an endless variety of content, such as movies, music, shows, comedies, so forth. One of the most important entertainment for people today is watching movies online. In fact, movie websites are not much on the Internet yet. The thesis seizes and develops based on this opportunity. The web-app is being integrated with popular technologies in the software market and the field of machine learning. It will bring to promote entertainment; this is also the aim of this thesis. These artificial intelligence technologies will help the system to handle complicated tasks. Also, the system will become powerful and ubiquitous with users. And, how the recommender system works for suggesting movies to the users, the engine will base on the popularities of the items, or our preference, interest. Finding the content for the user's favorite will be a challenging task for the system. According to the user's profile, the system will recommend suitable items for them. In addition, the thesis will focus on developing a single page application that includes the most popular technologies in the industry (such as Virtual DOM, JSX, Redux, Hook, Thunk, React, CSS Preprocessor, Django, so on) and the system is impossible without the recommendation engine, which also use the concepts such as tf-idf, vector space model, matrix factorization, singular value decomposition, cosine and so forth to analyst a dataset. Moreover, Single page application is a significant concept for developing the client-side to refresh the necessary pages and improve the user experience (UX), the user interaction. And one more technology is the encrypted user's account based on handling JSON Web Token (JWT). JWT will get a token from API of the web server after that store a token to LocalStoregae on the client-side. As a result, the performance on the server-side will be improved when the data is stored on the client-side. In addition, applying some models of recommendation systems is through the MovieLens dataset. The system will be scaled from 100,000 ratings to 10 million ratings in the future work and suggesting movies through the use of machine learning concepts (Content-Based Collaborative Filtering). These approaches will help constructing a model that predicts efficiently and accurately based on the user's profile when compared to the existing models. The objective of the thesis solves how to recommend accurately through models and catch up with the most recently technologies today.