Free University of Bozen-Bolzano
"Computing useful recommendations: still requires knowledge"
Abstract: Recommender systems have been introduced as information search and filtering tools for providing suggestions for items to be of use to a user. State of the art recommender systems mostly focus on the usage of data mining and information retrieval techniques to predict to what extent an item fits user needs and wants. But often they end up in making uninteresting suggestions especially in complex domains, such as tourism. In this talk, classical recommender systems ideas will be introduced and critically scrutinised in the attempt to better understand the role of observed and predicted choices and preferences. We will discuss some of the key ingredients necessary to build a useful recommender system. Hence, we will point out some limitations and open challenges for recommender systems research. We will also present a novel recommendation technique that leverages data collected from observation of tourists behaviour to generate more useful individual and group recommendations.
Biography: Francesco Ricci is full professor and dean of the Faculty of Computer Science, Free University of Bozen-Bolzano. F. Ricci has established in Bolzano a reference point for the research on Recommender Systems. He has co-edited the Recommender Systems Handbook (Springer 2011, 2015), and has been actively working in this community as President of the Steering Committee of the ACM conference on Recommender Systems (2007-2010). He was previously (from 2000 to 2006) senior researcher and the technical director of the eCommerce and Tourism Research Lab (eCTRL) at ITC-irst (Trento, Italy). From 1998 to 2000 he was system architect in the Research and Technology Department (Process and Reuse Technologies) of Sodalia s.p.a. F.Ricci has participated to several international research projects such as: RECOM (funded by Deutsche Telekom), etPackaging (funded by ECCA), European Tourist Destination Portal (funded by European Travel Commission), Harmoten (funded by IST), DieToRecs (Intelligent Recommendation for Tourist Destination Decision Making, funded by IST). Francesco Ricci is author of more than one hundred fifty refereed publications and, according to Google Scholar, has H-index 52 and around 17000 citations.
Robert Gordon University
"Learning to Compare with Few Data for Personalised Human Activity Recognition"
Abstract: Recent advances in meta-learning provides an interesting opportunity for CBR research, in similarity learning, case comparison and personalised recommendations. Rather than learning a single model for a specific task, meta-learners adopt a generalist view of learning-to-learn, such that models are rapidly transferable to related but different new tasks. Unlike task-specific model training; a meta-learner’s training instance, referred to as a meta-instance is a composite of two sets: a support set and a query set of instances. In our work, we introduce learning- to-learn personalised models from few data. We motivate our contribution through an application where personalisation plays an important role, mainly that of human activity recognition for self-management of chronic diseases. We extend the meta-instance creation process where random sampling of support and query sets is carried out on a reduced sample conditioned by a domain-specific attribute; namely the person or user, in order to create meta-instances for personalised HAR. Our meta-learning for personalisation is compared with several state-of-the-art meta-learning strategies: 1) matching network (MN) which learns an embedding for a metric function; 2) relation network (RN) that learns to predict similarity between paired instances; and 3) MAML, a model agnostic machine learning algorithm that optimizes the model parameters for rapid adaptation. Results confirm that personalised meta-learning significantly improves performance over non-personalised meta-learners.
Biography: Nirmalie’s research interests include both theoretical and practical aspects of machine learning and intelligent systems with particular focus on Case-based Reasoning (CBR), Text Mining and Machine Learning. Her current application focus is on digital health and algorithms for self-management and intervention.
The underlying aim of her research is to reduce the demand on manual intervention through the development of knowledge-rich representations for intelligent monitoring and decision support systems that reason with data, text and multimedia content. A common theme applicable across all content is to exploit similarity knowledge for effective modelling and semantic indexing.
She is an executive member of the British Computer Society’s Specialist Group on AI since 2002 and was editor of their Expert Update magazine (2005-2010). She has organised many research workshops on textual CBR, web CBD, Social Media Mining and more recently on Evolving systems and Deep Learning. She was also chair of the international CBR conference in 2011. Externally, Nirmalie has been a member of the programme committees of ECCBR, ICCBR, ECML/PKDD, IJCAI and FLAIRS. She also pioneers the cause for women in computing and co-chaired the Scottish women in computing research and networking 2015 event in Aberdeen. She currently has several PhD and funded research projects related to knowledge modelling and representation from time-series and social media data.