Today consumers are confronted with a very large number of products and services to choose from. This makes it difficult for users to find relevant products among a huge number of alternatives. A recommendation system is an extensive class of web applications that involves predicting the user responses to the options and helps users to find products of interest by analyzing their past transactions such as product views and purchases. There is also a similar problem in the real estate industry where thousands of properties are available for rentals or sales. In this work we firstly presented the details of a real estate recommender system developed for Zingat.com and then, we explained how we implemented a fully functional recommendation system for property listings in Turkey. Since the number of listings is huge and new listings come and go frequently, it is a challenge to build a successful recommender system. We tackled this challenge by building a system which uses collaborative filtering and content-based filtering, separately. We also designed a scalable system architecture which can function under heavy load. In the future we plan to further improve this system by using diversification techniques and new solutions.
Mathematics Subject Classification: 97R40
Keywords: Artificial intelligence, machine learning, real estate, recommendation engine, zingat.com