Product Recommendations through Neo4j by Analyzing Patterns in Customer Purchases

Recommendation system grows more important each day as user interaction on the internet grows in size and complexity. To achieve better user experience and personalized choice of products for each user, it is important to create a recommendation system that takes all the interaction of a user on the internet and analyzes it thoroughly to get a better understanding of the user. Understanding the user will benefit the business more as each user will get a personalized experience based on how they act. This study focuses on utilizing the graph database to gain insight into the user behavior and develop a recommendation system based on how users act on the internet. The recommender system will use the Neo4j database as it provides much functionality to work with, such as the Graph Data Science library and the Jaccard Similarity method. Using all the graph technologies that exist today, this study will enable businesses to give a personalized experience to user by providing a detailed, accurate, effective, and efficient recommendation to user.

随着互联网用户交互行为的规模持续扩张、复杂度不断提升，推荐系统的重要性亦与日俱增。为优化用户体验，为每位用户提供个性化的产品选购服务，构建一套能够全面采集并深入剖析用户在互联网上的全部交互行为、进而精准刻画用户画像的推荐系统，已然成为关键之举。对用户的精准刻画将为企业创造更多价值，因为每位用户都能获得与其行为模式高度匹配的个性化使用体验。本研究聚焦于借助图数据库（graph database）挖掘用户行为洞察，并基于用户在互联网上的行为模式开发一款推荐系统。本推荐系统将采用Neo4j数据库，因其内置丰富的可用功能，例如图数据科学库（Graph Data Science library）与杰卡德相似度（Jaccard Similarity）算法。依托当前主流的各类图技术，本研究将助力企业为用户提供细致入微、精准可靠、高效优质的个性化推荐服务，从而为每位用户打造贴合其需求的专属使用体验。