Explainable Recommender Systems with Knowledge Graphs and Language Models (Half-day)
In this tutorial, we cover the recent advances on explainable information retrieval through knowledge graphs (KGs), with recommender systems as the applicative scenario. We will start by highlighting the fundamental principles that drive the growing reliance on KGs in today’s user modelling, including relevant academic sources and real-world examples. Additionally, we will touch upon the recent techniques for explainable recommendation that combine language models and KGs. For a better understanding, we will present a taxonomy encompassing data types, methodologies, and evaluation metrics from existing research, designed for developing explainable recommender systems grounded in KGs. These conceptual elements will be interleaved by short hands-on sessions, providing attendees with implementations based on open-source tools and public data sets. Finally, we will show a comprehensive example of how the entire pipeline can be applied within an emerging domain, namely education. The tutorial will wrap up with an analysis of emerging issues and prospective trajectories in this domain.
- Giacomo Balloccu, University of Cagliari, Italy
- Ludovico Boratto, University of Cagliari, Italy
- Gianni Fenu, University of Cagliari, Italy
- Francesca Maridina Malloci, University of Cagliari, Italy
- Mirko Marras, University of Cagliari, Italy
Transformers for Sequential Recommendation (Half-day)
In this half-day tutorial, we draw parallels between language representation and sequential recommendation and show how large-scale sequential recommendation can be effectively addressed using the Transformer architecture. We start by describing the Transformer architecture and classical Transformer-based recommendation models, including SASRec and BERT4Rec. We then discuss the training objectives for training a Transformer-based recommender system, particularly showing how introducing recency during training helps the Transformer to learn quickly. Next, we consider large item catalogues, addressing first the interplay between loss functions and negative sampling for large numbers of items and, finally, how we may break item representations down into “sub-items” to allow efficient representation of large item catalogues by the Transformer model. We then describe the modern Generative approach to recommender systems and discuss how it can be used for complex recommendation goals. Finally, we discuss the role of transformer-based large language models in the future of recommender systems, describe current early advancements in the field, and discuss our vision of the future of recommender systems. The tutorial is based on classic papers in the domains of language modelling and sequential recommender systems, as well as our line of work on scaling transformers to large catalogues. Our target audience is both academic researchers and industry practitioners.
- Aleksandr V. Petrov, University of Glasgow, Scotland
- Craig Macdonald, University of Glasgow, Scotland
Affective Computing for Social Good Applications: Current Advances, Gaps and Opportunities in Conversational Setting (Half-day)
Affective computing involves examining and advancing systems and devices capable of identifying, comprehending, processing, and emulating human emotions, sentiment, politeness and personality characteristics. This is an ever-expanding multidisciplinary domain that investigates how technology can contribute to the comprehension of human affect, how affect can influence interactions between humans and machines, how systems can be engineered to harness affect for enhanced capabilities, and how integrating affective strategies can revolutionize interactions between humans and machines. Recognizing the fact that affective computing encompasses disciplines such as computer science, psychology, and cognitive science, this tutorial aims to delve into the historical underpinnings and overarching objectives of affective computing, explore various approaches for affect detection and generation, its practical applications across diverse areas, including but not limited to social good (like persuasion, therapy and support, etc.), address ethical concerns, and outline potential future directions.
- Priyanshu Priya, Indian Institute of Technology Patna, India
- Mauajama Firdaus, University of Alberta, Canada
- Gopendra Vikram Singh, Indian Institute of Technology Patna, India
- Asif Ekbal, Indian Institute of Technology Patna, India.
Quantum Computing for Information Retrieval and Recommender Systems (Half-day)
Quantum Computing (QC) is a research field that has been in the limelight in recent years. In fact, many researchers and practitioners believe that it can provide benefits in terms of efficiency and effectiveness when employed to solve certain computationally intensive tasks that may require years of high-performance computers. In Information Retrieval (IR) and Recommender Systems (RS) we are required to process very large and heterogeneous datasets by means of complex operations, it is natural therefore to wonder whether QC could also be applied to boost their performance. The goal of this tutorial is to show how QC works to an audience that is not familiar with the technology, as well as how to apply the QC paradigm of Quantum Annealing (QA) to solve practical problems that are currently faced by IR and RS systems. During the tutorial the participants will be provided with the fundamentals required to understand QC and to apply it in practice by using a real D-Wave quantum annealer through APIs.
- Maurizio Ferrari Dacrema, Polytechnic of Milan, Italy
- Paolo Cremonesi, Polytechnic of Milan, Italy
- Andrea Pasin, University of Padua, Italy
- Nicola Ferro, University of Padua, Italy
Recent Advances in Generative Information Retrieval (Half-day)
Generative retrieval (GR) has become a highly active area of information retrieval that has witnessed significant growth recently. Compared to the traditional ”index-retrieve-then-rank” pipeline, the GR paradigm aims to consolidate all information within a corpus into a single model. Typically, a sequence-to-sequence model is trained to directly map a query to its relevant document identifiers (i.e., docids). This tutorial offers an introduction to the core concepts of the novel GR paradigm and a comprehensive overview of recent advances in its foundations and applications. We start by providing preliminary information covering foundational aspects and problem formulations of GR. Then, our focus shifts towards recent progress in docid design, training approaches, inference strategies, and the applications of GR. We end by outlining remaining challenges and issuing a call for future GR research. This tutorial is intended to be beneficial to both researchers and industry practitioners interested in developing novel GR solutions or applying them in real-world scenarios.
- Yubao Tang, University of Chinese Academy of Sciences, China
- Ruqing Zhang, University of Chinese Academy of Sciences, China
- Zhaochun Ren, Leiden University, The Netherlands
- Jiafeng Guo, University of Chinese Academy of Sciences, China
- Maarten de Rijke, University of Amsterdam, The Netherlands
PhD Candidacy: A Tutorial on Overcoming Challenges and Achieving Success (Full day)
Undertaking a PhD is a demanding yet rewarding experience. PhD candidates develop a deep understanding of their research topic and acquire a wide range of skills, including (i) formulating research questions; (ii) conducting research ethically and rigorously; (iii) communicating research findings effectively to both academic and non-academic audiences alike; (iv) forging a profile as an independent researcher; and (v) developing a teaching portfolio. PhD candidates inevitably experience challenges during their candidature. These challenges can be overcome by applying various techniques to adapt and learn from these experiences. This tutorial will introduce attendees to several strategies to help them advance in the PhD process. It will be presented by two early career researchers in information retrieval, who have the unique perspective of being close enough to their time as PhD candidates to remember the highs and lows of PhD life yet far enough removed from the process to reflect on their experiences and provide insights. The tutorial will empower attendees to share their tips and tricks, review their practices for success, and refine the best productivity methods for them. It will provide a neutral platform for an open and honest discussion about the journey of undertaking a PhD, led by the presenters without judgement. This tutorial is intended for research students in all fields, especially those interested in learning from the experiences of others and developing strategies for success.
- Johanne Trippas, RMIT Univerity, Australia
- David Maxwell, Booking.com, The Netherlands
Query Performance Prediction: From Fundamentals to Advanced Techniques (Half-day)
Query performance prediction (QPP) is a core task in information retrieval (IR) that aims at predicting the retrieval quality for a given query without relevance judgments. QPP has been investigated for decades and has witnessed a surge in research activity in recent years; QPP has been shown to benefit various aspects, e.g., improving retrieval effectiveness by selecting the most effective ranking function per query. Despite its importance, there is no recent tutorial to provide a comprehensive overview of QPP techniques in the era of pre-trained/large language models or in the scenario of emerging conversational search (CS); moreover, while research in QPP has yielded promising results, its practical implementation and integration into real-world search engines remain a challenge.
In this tutorial, we have three main objectives. First, we aim to disseminate the latest advancements in QPP to the IR community. Second, we go beyond investigating QPP in ad-hoc search and cover QPP for CS. Third, the tutorial offers a unique opportunity to bridge the gap between theory and practice; we aim to equip participants with the essential skills and insights needed to navigate the evolving landscape of QPP, ultimately benefiting both researchers and practitioners in the field of IR and encouraging them to work around the future avenues on QPP.
- Negar Arabzadeh, University of Waterloo, Canada
- Chuan Meng, University of Amsterdam, The Netherlands
- Mohammad Alliannejadi, University of Amsterdam, The Netherlands
- Ebrahim Bagheri, Toronto Metropolitan University, Canada