Vasilis Efthymiou: Guided Data Analysis for Conversational Business Intelligence
When: June 15, 2022, 1.00pm-2.00pm (Helsinki time)
Where: Online via Zoom – https://tuni.zoom.us/j/68214190915
Abstract: In this talk, we will go over BI-REC, a conversational recommendation system for Business Intelligence (BI) applications. We will define the space of data analysis in terms of BI patterns, augmented with rich semantic information extracted from an OLAP cube definition, and we will see graph representations for analysis states. The BI-REC approach enables training accurate prediction models with little training data, achieves high-quality recommendations in terms of diversity and surprisingness among the predicted queries and has low prediction latency.
Speaker Info: Vasilis is a postdoctoral researcher at FORTH-ICS and a visiting faculty at University of Crete, Greece. Before that, he was a postdoctoral researcher at the database group of IBM Research – Almaden, working in the areas of entity matching, knowledge management and query answering, and an intern at IBM Research – Yorktown Heights, working on matching Web tables to Knowledge Graphs. He received his Ph.D. from University of Crete for research on “entity resolution in the Web of Data”, which is also the title of a book that he has co-authored.
Otto Sahlgren: Conceptualizing Fairness in Computational Team-Assembly
When: May 13, 1.00pm-2.00pm (Helsinki time)
Where: Online via Zoom – https://tuni.zoom.us/j/63180294060
Abstract: Software is increasingly used to assemble teams for various purposes and with different computational techniques. Despite the intuitive notion that team-membership is often a good to be distributed fairly, bias and fairness are topics that are somewhat neglected in work on computational team-assembly. I’ll motivate these topics and highlight sources for discriminatory bias in computational team-assembly pipelines. I’ll also discuss operationalizations of distributive fairness as a value in computational team-assembly. Finally, I argue that there are ethical and empirical reasons for adopting Representational Parity, a family of fairness targets grounded in the notion of equity, as an aim for computational team-assembly.
Speaker info: Otto Sahlgren is a Doctoral Researcher in Philosophy at Tampere University, whose work concerns the philosophical underpinnings of algorithmic fairness.
Ioannis Partalas: Learning Representations for Hotel Ranking
When: April 8, 2022, 12.00pm-1.00pm (Helsinki time)
Where: Online via Zoom – https://tuni.zoom.us/j/68781204125
Abstract: In this talk, Ioannis will present work on learning item representations from user click-sessions in the hospitality domain and more specifically from the Expedia Group online platforms. He will present in details the proposed neural architecture that leverages side information of items, like attributes and geographic information, in order to learn a joint embedding. Ioannnis will also explain how it addresses the cold-start problem which is typical in recommendation systems. Results in a downstream task show that including such structured information improves predictive performance. Finally, he will show through the results of on-line controlled tests that the model generates high quality representations that boosts the performance of a hotel recommendation system on Expedia travel platform.
Short CV: Ioannis Partalas works as Principal Machine Learning Scientist at Expedia Group. His current focus is learning representation in the context of recommendation and ranking systems. Previously, he worked as a Research Scientist in Viseo Group, France, on Natural Language Processing building scalable approaches for various tasks, such as text classification, named-entity recognition and opinion mining. Before that he was an associate researcher in Grenoble-Alpes University working on large-scale/extreme classification systems.
The aim of the RecSys Talks is to give the latest updates mostly on topics related to recommender systems, initiate discussions around them and form collaborations on topics related to the interests of the group.
The meeting is running once per month with presentations from the RecSys Group in Tampere University as well as from our colleagues outside the group.
Please send us an email at konstantinos.stefanidis@tuni.fi to inform us if you want to be reminded about the talks closer to their dates.