Faculty of Informatics and Statistics, Department of Information and Knowledge Engineering (DIKE)

Date and time: March 13 2014 (10:30 – 12:00).

Room: 336 RB Non–standard venue!


Recommender Systems for E-commerce and the MODGEN Recommendation Platform


  • Tomáš Řehořek, FIT ČVUT Praha

In the past decade, alongside with rapid expansion of E-commerce systems, there has been an enormous increase in the number of available products and services being offered to the users. In such a huge amount of content, it becomes more and more important to explore new ways how to organize the products, making it easy for the users to quickly find what they want. Our research focuses on a type of machine learning methods referred to as the Recommender Systems (RSs). Based on the past interactions of the users with such a system, RS is able to recommend previously unseen items being most likely relevant to a given user. Focusing particularly on methods known as the Collaborative Filtering (CF), we research algorithms capable of discovering patterns in global user behavior. This includes computing user-user and item-item similarities, discovering association rules between the products, or using even more advanced techniques such as the matrix factorization. We will briefly introduce our research, same as our effort to put RSs and CF into everyday practice, including live demo of the MODGEN Recommendation cloud service we are developing commercially. We’ll show visualizations of some real-world data, and demonstrate advanced features such as attribute-based filtering/boosting, allowing service operators to adjust the recommendations to better fit their business needs. Future is now, and for E-commerce and online marketing, RSs are the essential premise of not missing it!

Downloads: slides 1 

Powered by Resource Description Framework (RDF)