A Multimodal AI Assistant for Movie Discovery: Integrating Image Recognition into Movie Ticketing Systems

Tóm tắt

In the current context of digital transformation, enhancing user experience through artificial intelligence has become a critical factor for success in the entertainment and cinema industries. This paper presents CineJoy, a comprehensive movie ticket booking system featuring an advanced multi-modal AI assistant designed to create a seamless and intuitive movie discovery journey. Users can interact with the system not only through traditional text-based queries but also via image uploads, such as movie posters, promotional images, or scenes from a film. The system's core innovation is a two-stage retrieval architecture that combines AI-powered recognition with intelligent backend logic. Initially, the platform leverages Google's Gemini Vision API for multi-modal analysis to identify a movie title from the user-provided image. To address potential inaccuracies from AI recognition, a robust fuzzy matching algorithm is implemented on the backend. This algorithm employs techniques such as keyword extraction, Levenshtein distance, and a weighted scoring system to intelligently and accurately map the recognized name to the corresponding movie within the database. Upon successful identification, the AI assistant presents detailed movie information, available showtimes, and context-aware conversational prompts to enhance user engagement and guide them through the booking process. The CineJoy platform is developed using React (Vite) for the frontend, with a backend powered by NodeJS and ExpressJS, and MongoDB for data storage. User authentication and data protection are secured using JWT (JSON Web Tokens). Future work will focus on expanding the AI's capabilities to include deep personalization based on user profiles and action-oriented tasks for post-booking customer support.