We look forward to welcoming you to the 11th RecSys Amsterdam meetup. Hosted by SPARQUE, Utrecht with two interesting talks from industry and academia.
Doors open at 18:00h, talks start at 18:30h.
Drinks and pizza will be provided.
1. Chi Shing Chang, SPARQUE, "Managing evolving and sprawling recommendations models"
What are good recommendations? Data, domain experts and users may not necessarily be in agreement. Recommendations may best be based on methods that integrate the important aspects of all three sources. Warmteservice, a wholesaler of installation materials with physical branches and a webshop, needed high quality recommendations for complementary products. By presenting these recommendations, customers would not forget to order essential materials while Warmteservice sells more items. SPARQUE developed a model for Warmteservice that combines aspects from existing data, from experts and user behavior. I’ll talk about how the recommendations evolved and the tooling we used to manage the model and to give Warmteservice a performant, interactive and personalized recommendation engine.
Chi’s Bio: Chi Shing Chang is an ICT-entrepreneur. Co-founder of SPARQUE; a technology scale-up specialized in personalized E-commerce. Chi Shing is a graduate of the IMD MBA program and the Industrial Engineering program from University of Twente. He worked for several employers: Deli Maatschappij, Shell, Orange and McKinsey, before taking the leap to start a company. A generalist and an omnivore, he takes interest in a broad range of topics, including business, economics and human and organizational psychology.
2. Martijn Willemsen, Eindhoven University of Technology (TU/e), "Recommender systems to help people move forward"
Many real-life recommender systems are evaluated mostly on (implicit) behavioral data such as clicks streams and viewing times. However, such an approach has limitations and I will show how a user-centric approach can help better understand why users are satisfied or not, for example why users prefer diversification over prediction accuracy as it reduces choice difficulty. The behaviorist approach to evaluation also misses that users’ short term goals (i.e. their current behavior) might not be representative of the goals they want to attain (i.e. their desired behavior). What we like today might not be what we like tomorrow, because our tastes evolve or because our prior preferences and past behavior might not represent our current needs and goals. This is especially relevant in health and sustainability domains where people are in need of support while changing their current behavior. I will elaborate on an example from the energy recommendation domain, and show how a different type of recommender approach and interface might help users to save more energy. I will also discuss our recent work using Spotify to help people explore new tastes for unknown genres.
Dr. Martijn Willemsen is an expert on human decision making in interactive systems. He is working as an associate professor in the Human-Technology Interaction group of Eindhoven University of Technology (The Netherlands) and is the principle investigator of the recommender LAB at JADS Den Bosch (www.jads.nl). His primary interests lie in the understanding of cognitive processes of decision making by means of process tracing and in the application of decision making theory in interactive systems such as recommender systems. He is also an expert on user-centric evaluation of adaptive systems. He is part of the core team of the Customer Journey Research Program in the Data Science Center Eindhoven (DSC/e) and is teaching in the joint BSc and MSc data science programs of the Jheronimus Academy of Data Science (jads.nl). www.martijnwillemsen.nl
We look forward to welcoming you to the 10th RecSys Amsterdam meetup. Hosted by Persgroep with two interesting talks from industry and academia.
Doors open at 18:00h, talks start at 18:30h.
Drinks and pizza will be provided.
1. Daan Odijk, RTL, "Searching to be Entertained"
As the largest commercial broadcaster in a declining Dutch TV market, RTL is making a transition from a traditional TV company to a consumer-focused media company. RTL is embracing a closer relationship and more direct interaction with its viewers, followers and visitors. In this talk, I will share how we are using AI, recommender systems and search technology to help our users find the right content for them, ranging from the 1M daily visitors on our news website to the over 2B video plays we had in 2017, most of these on our rapidly growing video-on-demand platform Videoland.
Bio: Daan Odijk is the lead data scientist at RTL. In 2016, he obtained his PhD from the University of Amsterdam, researching search algorithms for news. Subsequently, he joined journalism start-up Blendle to lead the personalization team. At RTL since 2018, Daan leads a team of a dozen data scientists and engineers, delivering data-powered products across RTL, including personalization for RTL Nieuws and Videoland.
2. Julián Urbano, TU Delft. "The measure dilemma: which dataset-based measures are better to predict end-user satisfaction?"
The main goal of an evaluation experiment is to determine which systems perform well and which systems perform poorly on a task like retrieval or recommendation. However, there has been little systematic analysis regarding how well these evaluation results predict end-user satisfaction. For many researchers, reaching statistical significance is usually the objective, but not enough attention is paid to the real implications of the observed improvements. In this talk I'll present empirical results on the correlation between dataset-based results and user satisfaction in the task of music retrieval for recommendation. In particular, we'll discuss the effect that different annotation scales and measure formulations have on this correlation, and the implications for researchers, developers or reviewers.
Bio: Julián Urbano is an Assistant Professor at Delft University of Technology. His research is primarily concerned with evaluation in IR, working in both the music and text domains. Current topics of interest are the application of statistical methods for the construction of datasets, the reliability of evaluation experiments, statistical significance testing for IR, low-cost evaluation and stochastic simulation for evaluation. He has published over 50 research papers in related venues like Foundations and Trends in IR, the IR Journal, the Journal of Multimedia IR, SIGIR, ISMIR, CIKM, ICTIR and ECIR, winning two best paper awards and a best reviewer award.
We look forward to welcoming you to another RecSys Amsterdam meetup! Kindly hosted by Booking.com, we'll have three speakers in a mix from industry & academia. Doors open at 18:00, talks start at 18:30!
As usual, drinks and snacks will be provided.
1. Christophe van Gysel (UvA) will talk about work he's done at Bing: "Reply With: Proactive Recommendation of Email Attachments" (see: https://arxiv.org/abs/1710.06061)
2. Dung Chu (FD Mediagroep): "Recommendation systems at the FD"
Recommendation system is a popular data-driven tool used in many different business sectors. Within het FD we use an in-house recommendation engine in personalized email campaigns. In this talk, I will share our experience in: why we decided to develop our own recommendation engine; challenges that we encountered from business and technological points of view; first A/B testing results; and our future plan.
3. Themis Mavridis (Booking.com): "Learning To Match"
Booking.com is a virtual two-sided marketplace where guests and accommodation providers are the two distinct stakeholders. They meet to satisfy their respective and different goals. Guests want to be able to choose accommodations from a huge and diverse inventory, fast and reliably within their requirements and constraints. Accommodation providers desire to reach a reliable and large market that maximises their revenue. Finding the best accommodation for the guests, a problem typically addressed by the recommender systems community, and finding the best audience for the accommodation providers, are key pieces of a good platform. This work describes how Booking.com extends such approach, enabling the guests themselves to find the best accommodation by helping them to discover their needs and restrictions, what the market can actually offer, reinforcing good decisions, discouraging bad ones, etc. turning the platform into a decision process advisor, as opposed to a decision maker. Booking.com implements this idea with hundreds of Machine Learned Models, all of them validated through rigorous Randomised Controlled Experiments. We further elaborate on model types, techniques, methodological issues and challenges that we have faced. (see: https://export.arxiv.org/pdf/1802.03102)
We look forward to welcoming you to the 8th RecSys Amsterdam meetup. Hosted by FD Mediagroep, this media-themed meetup features four speakers from industry.
Doors open at 18:00h, talks start at 18:30h.
Drinks and snacks will be provided.
1. Ghida Ibrahim (Senior Data Scientist, (formerly at) Liberty Global (http://www.libertyglobal.com/)): "Recommender systems for video and TV products"
Ghida will give a talk about building recommender systems for video and TV products. Ghida's talk is inspired by her past experience working as a data scientist for Liberty Global. She will explain how recommender systems can be built using users viewership data and many variations of collaborative filtering, and how these can be assessed using A/B testing.
2. Bouke Huurnink and Roman Ivanov (XITE (https://www.xite.tv/)): "Music Video Recommendation@XITE"
Remember watching television and waiting for hours for that one music video to come on? Those days are long gone. With advanced music recommendation, smart TVs and ubiquitous screens, music television is transforming into an interactive, on-demand service. Giving you the music video you want, when and where you want it.
In this talk we discuss the evolution of music video recommendation at XITE. We will cover challenges that we have faced, our first recommender setup, and lessons learned/plans for the future.
3. Robbert van der Pluijm (Head of Bibblio Labs, Bibblio (http://www.bibblio.org/)): "Scaling a recommendation service - a threefold story"
Robbert works for Bibblio, which offers a content recommendation service for publishers. In this talk he'll share how they 1) built a local popularity recommender, 2) solved the challenges of scale and catalogue updating (for now) and 3) started to venture into deep learning.
PS: Take the entrance of the FD Building on the Wibautstraat side (https://email@example.com,4.917869,3a,75y,238.4h,95.88t/data=!3m6!1e1!3m4!1sIM7stUICPvldtSkhendu6w!2e0!7i13312!8i6656?hl=en) (There's another entrance at the Prins Bernhardplein, but it's closed after 6pm...).
In the next Recommender Systems Amsterdam - the first organised on meetup.com - we will have a combination of academic and industry talks. Nava Tintarev from TU Delft will describe interaction paradigms and explanation methods for recommender systems, Daan Odijk will discuss personalisation at Blendle, and Barend Linders will discuss online recommendation at the Dutch public broadcast, the NPO.
Doors open: 18.00. XITE will provide snacks and drinks.
Talks start: 18:30.
1. Nava Tintarev, 'Explain yourself! Arguing with Recommender Systems'
Nava is is an Assistant Professor and Delft Technology Fellow in the Web Information Systems group, Faculty of Electrical Engineering, Mathematics and Computer Science at TU Delft.
The complexities of many advice-giving systems often lead to people struggling to establish why a system chose what it did, to identify which alternatives were considered, and to determine why these alternatives were not selected or suggested. In other words, such systems are opaque, and a human (and particularly a non-expert) often struggles to understand their reasoning. During her talk Nava will introduce interaction paradigms, and methods for generating explanations (text and graphics) for recommender systems. She will also address how explanations can be designed to not only improve trust and transparency, but also improve the discovery of novel content and help users identify their own blindspots.
2. Daan Odijk, Blendle, Real-time Recommendations for News
Every morning, at Blendle, we have a huge cold-start problem when over 8.000 new articles from the latest newspapers arrive in our system. These articles are read by virtually no-one yet when we generate personalized newsletters for over a million users. We can thus not rely on collaborative filtering, nor can we use the popularity of the articles as clues for what our user might want to read. We overcome our cold-start problem by a mix of curation by our editorial team and automated content analysis using enrichments such as named entities, semantic links, authors, the language and plenty of stylometrics. Our editorial team get up at around 5am and is done reading and recommending their selection of articles around 7am, which is also the time we would ideally send out the newsletter. Starting a batch process only then would mean a prohibitively long delay. In this talk, I will outline our solution for real-time recommendations to address both challenges, based on a streaming infrastructure with Kafka at the core.
3. Barend Linders and Robbert van Waardhuizen, NPO, A sneak peek into the recommender system of the NPO
At the Dutch Public Broadcasting Organisation (NPO) the Marketing Intelligence team is responsible for producing the recommendations served on the online portal (npo.nl (http://npo.nl/) and corresponding apps). This is currently done via collaborative filtering on a series level (a stream or episode is part of a series). By using the instances of what content is watched together by users we calculate what content is expected to match what is currently being watched. I will discuss some of the limitations of this approach and future improvements we would like to make, as it is a continuous work in progress. Finally we will discuss current explorations on how to make our recommendations more diverse. Since it is part of our mission to ‘connect and enrich the Dutch audience…’ we want to showcase the immense diversity present in our video content.
Barend is a Hydrogeologist turned Data Scientist currently working within the Marketing Intelligence team at the NPO. Robbert van Waardhuizen is currently doing his Data Science Msc. thesis research at the NPO and will talk about his improvements on the algorithm to suggest more diverse recommendations.