We also show how we have used this technology to build MindReader, a recommendation system using graph technologies (explained later in this article) allowing users to collaboratively build a dataset unlike any other dataset used in the research field of personalized recommendation. While many recommender systems rely on several subsystems interacting with each other (e.g., machine learning clusters training and pulling data from a central database), we will implement a recommender that runs directly on the database itself - and very efficiently so - by exploiting the expressive power of Knowledge Graphs. In this article, we will go through how we can build an effective recommendation system using only Neo4j. Their purpose is simple: recommend the items/movies/people that a specific user will most likely buy/watch/become friends with. Recommendation systems, or recommenders, are used by a huge number of platforms including Amazon, Netflix, Facebook and many other e-commerce and service provision platforms. Based on what you have watched and rated, it builds a profile of your tastes in terms of genres, plots, actors and more, and uses this profile to recommend movies that fit to your taste. Netflix uses a powerful recommendation system to generate this list. Some new releases, some popular among other users, and most interestingly, some Top Picks for You. When you visit Netflix, you are met by several lists of movies for you to watch.
0 Comments
Leave a Reply. |