ADS Drinks & Data: Machine Learning & Recommendation Systems

Join us on Wednesday, November 23rd at 16:00!

Recommender systems are an important class of machine learning algorithms that offer "relevant" suggestions to users. Categorized as either collaborative filtering or a content-based system, we will be talking about different projects and perspectives with Evangelos Kanoulas (IvI, UvA), Makoto Miyazaki (Dataiku) and Marthe Möller (DSC, UvA).

Location
Science Park 904, 1098 XH Amsterdam
Room: A1.28 - When entering Science Park 904, walk up staircase on the left to first floor and then all the way right and room is at the end of the corridor. There are signs pointing to the room above.

Programme
15:55 Walk-in
16:00 Introduction & Welcome
16:05 Talk #1: Evangelos Kanoulas (IvI, UvA)
16:20 Discussion
16:25 Talk #2: Makoto Miyazaki (Dataiku)
16:40 Discussion
16:45 Talk #3: Marthe Möller (DSC, UvA)
17:00 Discussion
17:30 End and Networking!

Talk #1 by Evangelos Kanoulas
Recommender systems are intelligent systems that are able to personalise recommendations on the basis of your previous interactions with the system, but also on the basis of other users of the system who share similar interests and wills. Recommender systems allow users to find things of their interest (products, movies, songs, etc.) within an ocean of possibilities and they have shown impressive performance. What if however you do not like what is recommended to you? Can you control in a comprehensive way what is suggested to you? In this talk I will discuss two research directions towards controllable recommender systems, (a) conversational recommenders that are able to explicitly ask questions to the users before recommending items to them, and (b) disentanglement for recommendation that allow users to actively change recommendations based on their interests.

Talk #2 by Makoto Miyazaki
Ramen is a Japanese delicacy but has become pretty much an international thing. Here in the Netherlands as well, you must have your favorite ramen restaurant. What if I can suggest to you a ramen restaurant in Tokyo that best matches your taste? In this presentation I will introduce a recommendation system of the ramen restaurant based on the analysis of customer reviews using NLP technique.

Talk #3 by Marthe Möller
Amongst others, communication scholars study social media comments to learn more about computer-mediated communication. Whereas previous research has developed algorithms that automatically detect spam among social media comments, such algorithms are not always suitable to select comments that are relevant for the projects of communication scholars. Therefore, the present project investigates how supervised machine learning can be used to detect those social media comments that communication scholars can use to advance their theoretical understanding of the antecedents and consequences of digital media usage. In doing so, it discusses the various decisions that scholars need to make when using supervised machine learning and the consequences that these decisions have for the results generated by supervised machine learning models.