Dr. Sylvain Castagnos

Research

Interests

Through numerous research programs, competitions, and economic surveys, automated information retrieval systems and recommender systems have been proven to be efficient and useful by reducing the cognitive load and time required during the search and access to data. Over the past two decades of research within this field, this improvement of human-computer interactions is mainly relying on increasing systems' accuracy at different levels. Usage mining techniques aim at inferring accurate preferences, habits, and interests and building profiles from users' actions. Collaborative and content-based filtering make use of these profiles to provide users with relevant recommendations. Ontology-based systems formally define concepts within a domain, thus reducing ambiguity.

All these machine learning models and algorithms are evaluated relatively to true risk and empirical risk, leading to very accurate contents. Yet, a crucial aspect is missing within these evaluation metrics. It does not take into account human factors playing a role within the decision process. Even the most relevant information is not sufficient to maximize users' acceptance/adoption rate, and satisfaction. The time has come to design holistic intelligent systems that provide the right information at the right time, in the correct manner, in agreement with users' policy and with valuable arguments.

Thus, my research challenges consist in: (1) identifying human factors that play a role within decision making an/or maximize users' acceptance, adoption and satisfaction, (2) integrating these factors in machine learning algorithms, (3) designing interfaces to improve human-computer interactions.

My topics of interest include (but are not limited to) the following:

Human factors and decision making (diversity, memory, personality, emotions, mood, culture, ...)
Preserving privacy while modelling users and recommending items
Trust and reputation
Context and information retrieval
Impact of recommenders on decision process
Multi-criteria optimization (privacy vs. accuracy, diversity vs. similarity, scalability vs. time constraints, ...)
User studies (identifying human factors, evaluation of recommender systems)
Social influence (leaders, explicit and implicit social networks, maximizing acceptance, manipulation...)
User-centered design and adaptation of interfaces
Presentation and explanations in recommender systems
Visual representation of data

I have six application domains: e-commerce, web applications and services (intranet, online music services and movie RS), e-learning, e-health, cultural heritage, and video games.

Keywords: diversity, holistic modeling, cognitive sciences, distributed collaborative filtering, sequence-aware RS, clustering methods, markov models, deep learning, natural language processing, unsupervised learning, supervised learning, reinforcement learning, bayesian networks, implicit and explicit modeling of preferences and usages, preference elicitation, human computer interaction, artificial intelligence, chaos theory, intelligent user interfaces, constraint satisfaction problem solving, web information systems, intelligent agents, case-based reasoning, data mining, information retrieval.

Main Results

Modeling Memory and Inhibition while Acquiring new Skills
Guilherme Medeiros Machado, Geoffray Bonnin, Sylvain Castagnos, Lara Hoareau, Aude Thomas and Youssef Tazouti. An Approach to Model Children's Inhibition during Early Literacy and Numeracy Acquisition (pdf). In 21st International Conference on Artificial Intelligence in Education (AIED 2020), Cyberspace, July 2020.

Early literacy and numeracy skills are developed during childhood at kindergarten level. Among the many factors that influence the development of such skills, the literature shows that the executive function of inhibition - i.e. the blocking out or tuning out of information or action that is irrelevant to the learning task - is one of the most important. There are many tests to assess children's inhibition skills; however, such tests are generally time-consuming and have a short lifespan. In this context, we propose a computational approach to model children's inhibition skills by using only student traces from a learning app as input. We propose a mathematical formalization of three related inhibition features, which could be used as input to classification algorithms.
Multi-Objective Sequence-Based Recommender System
Pierre-Edouard Osche, Sylvain Castagnos and Anne Boyer. AntRS: Recommending Lists through a Multi-Objective Ant Colony System (pdf). In 41st European Conference on Information Retrieval (ECIR 2019)Cologne, Germany, April 2019.

When people use recommender systems, they generally expect coherent lists of items. Depending on the application domain, it can be a playlist of songs they are likely to enjoy in their favorite online music service, a set of educational resources to acquire new competencies through an intelligent tutoring system, or a sequence of exhibits to discover from an adaptive mobile museum guide. To make these lists coherent from the users' perspective, recommendations must find the best compromise between multiple objectives (best possible precision, need for diversity and novelty). We propose to achieve that goal through a multi-agent recommender system, called AntRS. We evaluated our approach with a music dataset with about 500 users and more than 13,000 sessions. The experiments show that we obtain good results as regards to precision, novelty and coverage in comparison with typical state-of-the-art single and multi-objective algorithms.
Understanding User's Context through the Analysis of Diversity within Sequences of Consultations
Amaury L'Huillier, Sylvain Castagnos and Anne Boyer. Understanding Usages by Modeling Diversity over Time (pdf). In 22nd International Conference on User Modelling, Adaptation and Personalization (UMAP 2014), pp. 81-86, Aalborg, Denmark, July 2014.

Let's imagine a system that can recommend the kind of music (among other application domains) you like to listen when you are at work, without having to know your location, IP address or even to ask your current mood. In this paper, we bring this dream closer by proposing a model that can automatically understand the user's current context. This model, called DANCE, analyzes the attributes of the items in your recent history and monitors the relative diversity brought by your consultations over time. We validated our approach with a music corpus of 100 users and a global history of 204,758 plays.
Improving Accuracy with Association Rules in Recommender Systems
Sylvain Castagnos, Armelle Brun and Anne Boyer. Probabilistic Reinforcement Rules for Item-Based Recommender Systems (pdf). In 8th European Conference on Artificial Intelligence (ECAI 2008), pages 823-824, Patras, Greece, July 2008.

A Collaborative Filtering algorithm is broken down into 3 parts. Firstly, the system needs to collect data about all users under the form of explicit and/or implicit ratings. Secondly, this data is used to infer predictions, that is to say to estimate the votes that the active user would have assigned on unrated items. Finally, the recommender system suggests to the active user items with the highest estimated values. The aim of this work was to improve the second step for Item-Based systems where classes of similar items are built by computing each pairwise correlations. We assume that, in some cases, pairwise similarities may be insufficient to explain the interest of a user for an item. Guided by this hypothesis, we propose a new model, called RIBA, which evaluates similarities of triplets, rather than pairs of items. Our main contribution consists in automatically detecting non-trivial associations between items in order to make the recommender engine more accurate. These associations take the form of reinforcement rules. These rules are determined by exploiting probabilistic skewnesses in triplets of items, and are then used to refine predictions of the system. Our approach only requires explicit ratings from users and doesn't need any information about items. Our model shows an improvement from 6 to 8% as regards the HMAE accuracy measure.

Other Relevant Publications Sylvain Castagnos, Armelle Brun and Anne Boyer. Probabilistic Reinforcement Rules for Item-Based Recommender Systems (pdf). In 4th European Starting AI Researcher Symposium (STAIRS 2008), pages 36-46, Patras, Greece, July 2008.

Enhancing Privacy in Recommender Systems
Sylvain Castagnos and Anne Boyer. Privacy Concerns when Modeling Users in Collaborative Filtering Recommender Systems. Book Chapter in Social and Human Elements of Information Security: Emerging Trends and Countermeasures, IdeaGroup,Inc. September 2008.

This chapter investigates ways to deal with privacy rules when modeling preferences of users in recommender systems based on collaborative filtering. It argues that it is possible to find a good compromise between quality of predictions and protection of personal data. Thus, it proposes a methodology that fulfills with strictest privacy laws for both centralized and distributed architectures. The authors hope that their attempts to provide an unified vision of privacy rules through the related works and a generic privacy-enhancing procedure will help researchers and practitioners to better take into account the ethical and juridical constraints as regards privacy protection when designing information systems.

Modeling Preferences and Providing Recommendations in Peer-to-Peer Architectures
Sylvain Castagnos and Anne Boyer. Personalized Communities in a Distributed Recommender System (pdf). In Proceedings of the 29^th European Conference on Information Retrieval (ECIR 2007), pages 343-355, Rome, Italy, April 2007 (acceptance rate 19%).

The amount of data exponentially increases in information systems and it becomes more and more difficult to extract the most relevant information within a very short time. Among others, collaborative filtering processes help users to find interesting items by modeling their preferences and by comparing them with users having the same tastes. Nevertheless, there are a lot of aspects to consider when implementing such a recommender system. The number of potential users and the confidential nature of some data are taken into account. This paper introduces a new distributed recommender system based on a user-based filtering algorithm. Our model has been transposed for Peer-to-Peer architectures. It has been especially designed to deal with problems of scalability and privacy. Moreover, it adapts its prediction computations to the density of the user neighborhood.

Other Relevant Publications
Sylvain Castagnos and Anne Boyer. Modeling Preferences in a Distributed Recommender System (pdf). In Proceedings of the 11^th International Conference on User Modeling (UM 2007), pages 400-404, Corfu, Greece, June 2007.

Making Collaborative Filtering Scalable for Client/Server Architectures
Sylvain Castagnos and Anne Boyer. A Client/Server User-Based Collaborative Filtering Algorithm: Model and Implementation (pdf). In proc. of the 17^th European Conference on Artificial Intelligence (ECAI 2006), in the 4^th Prestigious Applications of Intelligent Systems special section (PAIS), pages 617-621, Riva del Garda, Italy, August 2006 (nominated for the best paper award, acceptance rate 26%).

This paper describes a new way of implementing an intelligent web caching service, based on an analysis of usage. Since the cache size in software is limited, and the search for new information is time-consuming, it becomes interesting to automate the process of selecting the most relevant items for each user. We propose a new generic model based on a client/server collaborative filtering algorithm and a behavior modeling process. In order to highlight the benefits of our solution, we collaborated with a company called ASTRA which is specialized in satellite website broadcasting. ASTRA has finalized a system sponsored by advertisement and supplying to users a high bandwidth access to hundreds of websites for free. Our work has been implemented within its software architecture and, in particular, within its recommender system in order to improve the satisfaction of users. Our solution is particularly designed to address the issues of data sparsity, privacy and scalability. Because of the industrial context, we consider the situation where the set of users is relatively stable, whereas the set of items may vary considerably from an execution to another. In addition to the model and its implementation, we present a performance assessment of our technique in terms of computation time and prediction relevancy.

Projects

BOOM (ANR National Project, 2021-2025): Modeling and Opening Opinion Bubbles
Needle (University of Lorraine, AAP PRD): Contributory browser application for discovering and sharing online content
Music'Mouv (University of Lorraine, Projet Interdisciplinarité 2021): Exploitation of physiological and biomechanical data for the measurement and prediction of emotion when listening to music in order to facilitate the initiation of walking
MBANv2 (non-economic valuation project, 2019-2024): Virtual Museum of Fine Arts of Nancy
LINUMEN (eFran CDC Investissements d'avenir, 2017-2021): LIttératie et NUMératie Emergentes par le Numérique
METAL (eFran CDC Investissements d'avenir, 2016-2021): Modèles et Traces au service de l'Apprentissage des Langues
CROSSCULT (European Project H2020, 2016-2019): Empowering reuse of digital cultural heritage in context-aware crosscuts of European history
AMI (Coordinator, Region of Lorraine and FEDER, in collaboration with CHU Nancy, 2016-2020): Analyse de la Mémoire à travers les Interactions
Diversity and Context (Coordinator, Region of Lorraine and GrandNancy, in collaboration with Yupeek, 2014-2017): Modélisation et contrôle de la diversité dans les systèmes de recommandation au cours du temps
SCIARAT (CPER, 2016-2020): Stimulation Cognitive, Intelligence ambiante, robotique d'assistance et Télémédecine
EGL (PEPS Mirabelle, in collaboration with CREM, 2014-2015): Expressive Game Lab
Dr Sport (e-Health FWD, Région Lorraine, 2014-2015): Diagnostic probable de pathologies liées à des accidents du sport
Cultural impacts (Coordinator, STIC Asia, in collaboration with Hong Kong Baptist University and AIT, 2011-2012): Cultural Impacts on User Modeling and Interaction Design in Recommender Systems
ALGODEC (EPFL, European Project COST Action IC0602, 2008-2010): Decision Theory in Recommender Systems
Sat'n'Surf (ESA Consortium, in collaboration with CRPHT and SES ASTRA, 2004-2005): Agents collaboratifs pour un filtrage intelligent

Software

MBANv2 (Sylvain Castagnos, Florian Marchal, Yohann Fransot, Sophie Mouton, Sophie Toulouse, Charles Villeneuve de Janti, Michèle Leinen, Jean-Paul Darada, Gabriel Daubenfeld, Jérémy Germain, Julien Hans, Julie Cunin, Mathias Rihet, Titouan Boudard, Alexandre Remiatte, Juliette Kratz, Anthony Pellizzeri, Alix Delannoy, Violaine Ferrandez, Loïc Jimenez, Martin Lemaitre, Robin Sevillano, Alexandre Bertrand, Djalila Mahmoudi, Morgane Colle, Loane Didierjean, Kevin Degiorgio, Yann Richard)

The BIRD Team is currently developing a virtual online museum as an evaluation platform for sequence-based recommender systems (Licence CC-BY-NC-SA 4.0). See a video of presentation here.
KIWI Music (Myriam Delaruelle, Elise Richard, Joris Favier, Gaël Hopp, Michel-Ange Dagrain, Roderick Pierre, Stéphane Mazzei, Amaury L'Huillier, Sylvain Castagnos, Geoffray Bonnin)

The KIWI Team has developed an online music recommender system as an evaluation framework for recommender algorithms.
Dr Sport (Laura Infante Blanco, Sylvain Castagnos, Thierry Weizman)

Dr Sport is the first full service dedicated to diagnosis of sport pathologies (from the analysis of the pathology to the orientation towards the nearest competent professionals). It relies on Artificial Intelligence techniques and will be available on iOS, Android, and on a website.
A.M.E. (Florian Marchal, Sylvain Castagnos)

This software aims at analysing usages (movements, gaze data, clicks, timestamps, ...) while passing neuropsychological tests such as TMT, and to automatically detect cognitive and memory disorders. It is compatible with Leap Motion and Tobii X1 Light Eye-Tracker.
DANCE (Amaury L'Huillier, Sylvain Castagnos)

DANCE analyses the attributes of the items in users' recent history and monitors the relative diversity brought by their consultations over time.
DANCE Visualisator (Alexandre Kerangall, Amaury L'Huillier, Sylvain Castagnos)

This is a Java software that allows researchers to dynamically visualize results from the DANCE model (see above).
Precog (Benjamin Gras, Benoit Hamann, Sylvain Castagnos)

This software has been designed to verify the existence of a link between memory and gaze data. On one hand, we are starting a collaboration with the Nancy University Hospital to study memory troubles among elderly persons, with the goal of automatically diagnosing memory issues based on fixation points, saccades and scanpaths. On the other hand, we are working on predictive models to provide recommendations of educational resources in Intelligent Tutoring Systems, according to what the user looked at and what the system assumes to be memorized.
Concretely, Precog allows users to play to the well-known game called "Concentration" (also known as Memory, Pelmanism, Shinkei-suijaku, or Pexeso). In the meantime, the system collects gaze data such as fixation points and durations. Then, it uses these pieces of data to predict which items will be memorized according to primacy and recency effects. Our machine learning algorithm can still be improved, but first results are encouraging. Our application has been developed in C# so as to be compatible with a Tobii X1 Light eye-tracker and Tobii SDK. This prototype has been presented to more than 10,000 persons during the Renaissance Nancy 2013 exhibit. This software is available on demand.
StreetViewer (Florian Marchal, Amaury L'Huillier, Maxime Amblard, Sylvain Castagnos)

This software has been developed in C# for the need of the Renaissance Nancy 2013 exhibit. It has been designed to study and illustrate the added value of gesture-based software in certain contexts. StreetViewer allows users to use the Google service called StreetView without keyboard or mouse. A set of gestures have been defined and interpreted as actions by the software. It helped us to analyse users' behaviours and understanding of such a system. We plan to use this software within a research project around well-being and aging process. It is also available on demand.
Commun-!T (Gabin Personeni, Liang Qiao, Sylvain Castagnos)

This software, developped in PHP and Flex in 2012 is able to draw communities of users under the form of hyperbolic graphs. The input data are information about users and items. These data comes from various corpuses used and/or generated by KIWI algorithms. It is then able to display the links between users according to their preference similarities, their social relationship, or the level of trust they have in each others. It is also able to display the list of recommandations coming from and computed for each user.
FRAC+ Software

FRAC+ is a software developed in JAVA which fits with client/server architectures. Its goal is to provide a personalizing service within enterprise portals and e-commerce applications. The client part includes a module for user modeling and a recommender engine. The server part contains a clustering collaborative filtering algorithm allowing the system to build virtual communities of interests. This enables to acquire experience from similar population and to bypass the problems of missing data in individual preference models, thus improving the recommendation computations on the client side. This software has been successfully integrated in real industrial contexts as, for example: the Casablanca website broadcasting service of the company SES ASTRA (within the framework of the ESA Sat'n'Surf project), the intranet portal of the technological foresight department of Crédit Agricole S.A. (Paris, France), etc.
SofoS document sharing platform

SofoS is an experimental peer-to-peer documentary platform from the LORIA Laboratory allowing people to share items in various formats (textual, video, audio, websites, etc.) and to do some researches in order to find new contents available on the platform. This software has been developed in JAVA and relies on the JXTA open-source toolkit. SofoS includes a powerful recommender system being able to test and validate new filtering algorithms. This software already includes the highly-scalable AURA collaborative filtering algorithm which provides very accurate recommendations based on implicit and explicit criteria in real time.

Seminars

Invited Talk

When Diversity Is Needed. But Not Expected!, Seminar Complex Networks, LIP6. Paris, France, 17 Mai 2018.
Research and Access to Information on the Internet, 9^th french conference of Electronics (CNRS). Dijon, France, 5 Juin 2007.

Other Activities

Conference Reviews

Regular Program Committee Member of the International ACM Conference on User Modelling, Adaptation and Personalization (UMAP 2018, UMAP LBR 2018, UMAP 2019, UMAP LBR 2019, UMAP 2020, UMAP LBR 2020, UMAP 2021, UMAP 2022, UMAP LBR 2022, UMAP LBR 2023).
Program Committee Member of the ACM Web Conference (TheWebConf 2022, 2023).
Program Committee Member of the International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2022).
Regular Program Committee Member of the International ACM Conference on Recommender Systems RecSys (2013, 2015, 2016, 2017, 2018, 2020, 2021, 2023).
Regular Program Committee Member of the IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2018, 2019).
Reviewer for the French National Research Agency (ANR 2018, 2020).
Organization Committee and Program Committee Member of LSAC 2019. 3rd Annual Learning & Student Analytics Conference (Nancy, France)
Reviewer for the Journal "User Modeling and User-Adapted Interaction" (UMUAI) in 2018 and 2022.
Reviewer for the PeerJ Journal in 2022.
Organization Committee and Program Committee Member of MobileCH 2018. 2nd Workshop on Mobile Access to Cultural Heritage (Barcelona, Spain)
Organization Committee Member of the French Conference PFIA 2018. Plate-Forme en Intelligence Artificielle (Nancy, France).
Reviewer for the Journal: Information Processing and Management (1 paper). December 2017.
Regular Program Committee Member of the French National Conferences CORIA (2013, 2014, 2015, 2016). Member of the best paper committee at CORIA 2013.
Program Committee Member of the Doctorial Consortium track in the International Conference on Multimodal Interaction (ICMI 2014).
Organization Committee Member of the French Joint Conferences CORIA 2014 and CIFED 2014. Semaine du Document Numérique et de la Recherche d'Information (Nancy, France)
Organization of the special session on Human Factor in Information Retrieval (HFIR'12) of the ISMIS 2012 International Conference. Hong Kong, China.
Reviewer for the ACM Transactions on Intelligent Systems and Technology (TIST) and the International Journal of Information Technology (IJIT).
Program Committee Member of the International ACM Conference on Intelligent Interactive Technologies and Multimedia (IITM2010). Allahabad, India, December 2010.
Program Committee Member of the International ACM Conference on Intelligent User Interfaces (IUI 2010). Hong Kong, China, February 2010.
Reviewer of the 10^th ACM Conference on Electronic Commerce (2 papers). Stanford, California, June 2009.
Reviewer of the 26^th SIGCHI 2008 Conference (3 papers). Florence, Italy, April 2008.

Administrative and Technical Responsibilities

Member of the IDMC Council. From January 2019.
Member of the LORIA Laboratory Council. From September 2018.
Webmaster of the MobileCH 2018 conference website. From March 2018.
Webmaster of the PFIA 2018 conference website. From July 2017.
Member of the working group "ateliers des possibles", Métropole du GrandNancy. From September 2015.
Participation to Moments d'invention 2016, Métropole du GrandNancy.
Co-chief of the certificate of Computer Science and the Internet (c2i, level 1) at the University of Lorraine. From October 2011 to August 2018.
Co-organizer of the Annual Forums on Cognitive Sciences (from 2011). Department of Mathematics and Computer Sciences, University of Lorraine.
Member of the cluster Nancy Numérique. From September 2014.
Webmaster of the KIWI Team (LORIA). From January 2008.
Ex-head of the communication within the Department of Mathematics and Computer Science, University of Lorraine. July 2011 - August 2014.
Participation to Nuit des chercheurs 2014.
Participation to Renaissance Nancy 2013.
Past-member of the LORIA Laboratory Council. June 2007 - November 2008.
Ex-administrator of the Group HCI's Linux Server. November 2008 - July 2010.
Ex-webmaster of the MAIA Team (LORIA). December 2005 - December 2007.

Awards for the project "e-veille" (Sailendra S.A.S.)

Co-winner of the national competition for aid in innovative technology company creation (Category "Creation-Development"). June 2007.
Co-winner of the competition "Entreprendre 2007" (Category "Achievement"). June 2007.
Co-winner of the masters 2007 of company creation. December 2006.
Co-winner of the national competition for aid in innovative technology company creation (Category "Emergence"). June 2006.

Publications

International and National Journals

Modelling Children's Inhibitory Skills using Learning Data from an Educational App
Guilherme Medeiros Machado, Geoffray Bonnin, Sylvain Castagnos, Lara Hoareau, Aude Thomas and Youssef Tazouti
In Journal of Computer Assisted Learning, JCAL 2023
Learning Analytics Made in France: The METAL project
Armelle Brun, Geoffray Bonnin, Anne Boyer, Sylvain Castagnos and Azim Roussanaly
In International Journal of Information and Learning Technology, IJILT 2019
First Attempt to Predict User Memory from Gaze Data (extended version of our ICTAI 2016 paper, selected among the best papers of the conference)
Florian Marchal, Sylvain Castagnos and Anne Boyer
In International Journal on Artificial Intelligence Tools, IJAIT 2018
Modéliser la diversité au cours du temps pour détecter le contexte dans un service de musique en ligne
Amaury L'Huillier, Sylvain Castagnos and Anne Boyer
Revue des Sciences et Technologies de l'Information - Série TSI : Technique et Science Informatiques, Lavoisier, 2016
La diversité : entre besoin et méfiance dans les systèmes de recommandation (pdf)
Sylvain Castagnos, Armelle Brun and Anne Boyer
Interaction Intelligence Information (International Journal I3), 2014
From Community Detection to Mentor Selection in Rating-Free Collaborative Filtering (pdf)
Armelle Brun, Sylvain Castagnos and Anne Boyer
Advances in Multimedia, 2011

Book Chapters

Social recommendations : mentor and leader detection to alleviate the cold-start problem in collaborative Filtering (pdf)
Armelle Brun, Sylvain Castagnos and Anne Boyer
Book Chapter in Social Network Mining, Analysis and Research Trends : Techniques and Applications, 2011
Privacy Concerns when Modeling Users in Collaborative Filtering Recommender Systems (pdf)
Sylvain Castagnos and Anne Boyer
Book Chapter in Social and Human Elements of Information Security: Emerging Trends and Countermeasures, IdeaGroup,Inc. September 2008

Highly Selective Peer-Reviewed Conference Papers

A Multi-Factorial Analysis of Polarization on Social Media (pdf)
Celina Treuillier, Sylvain Castagnos and Armelle Brun
In proc. of the 31th ACM Conference on User Modeling, Adaptation and Personalization (ACM UMAP 2023 Late-Breaking Results). Limassol, Cyprus, June 2023.
Being Diverse is Not Enough: Rethinking Diversity Evaluation to Meet Challenges of News Recommender Systems (pdf) (slides)(video)
Celina Treuillier, Sylvain Castagnos, Evan Dufraisse and Armelle Brun
In proc. of the Workshop on Fairness in User Modeling, Adaptation and Personalization (FairUMAP 2022), in conjunction with the 30th ACM Conference on User Modeling, Adaptation and Personalization (ACM UMAP 2022). Barcelona, Spain, July 2022.
An Approach to Model Children's Inhibition during Early Literacy and Early Numeracy Acquisition (pdf)
Guilherme Medeiros Machado, Geoffray Bonnin, Sylvain Castagnos, Lara Hoareau, Aude Thomas and Youssef Tazouti
In proc. of the 21th International Conference on Artificial Intelligence in Education (ACM AIED 2020). ~~Ifrane, Morocco~~ Cyberspace, July 2020.
Eye Gaze Sequence Analysis to Model Memory in E-education
Maël Beuget, Sylvain Castagnos and Anne Boyer
In proc. of the 20th International Conference on Artificial Intelligence in Education (ACM AIED 2019). Chicago, USA, June 2019. (acceptance rate: 32%)
AntRS: Recommending Lists through a Multi-Objective Ant Colony System
Pierre-Edouard Osche, Sylvain Castagnos and Anne Boyer
In proc. of the 41st European Conference on Information Retrieval (ECIR 2019). Cologne, Germany, April 2019. (acceptance rate: 23%)
Are Item Attributes a Good Alternative to Context Elicitation in Recommender Systems?
Amaury L'Huillier, Sylvain Castagnos and Anne Boyer
In proc. of the 25th ACM International Conference on User Modelling, Adaptation and Personalization (UMAP 2017). Bratislava, Slovakia, July 2017.
A First Step toward Recommendations Based on the Memory of Users
Florian Marchal, Sylvain Castagnos and Anne Boyer
In proc. of the 28th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2016). San Jose, California, USA, November 2016.
The New Challenges when Modeling Context through Diversity over Time in Recommender Systems
Amaury L'Huillier, Sylvain Castagnos and Anne Boyer
In Proc. of the 24th Conference on User Modelling, Adaptation and Personalization (UMAP 2016 Doctoral Consortium). Halifax, Canada, July 2016.
Tell Me What You See, I Will Tell You What You Remember
Florian Marchal, Sylvain Castagnos and Anne Boyer
In Proc. of the 24th Conference on User Modelling, Adaptation and Personalization (UMAP 2016). Halifax, Canada, July 2016. (acceptance rate: 41%)
Toward a Robust Diversity-Based Model to Detect Changes of Context (pdf)
Sylvain Castagnos, Amaury L'Huillier and Anne Boyer
In Proc. of the 27th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2015). Vietri sul Mare, Italy, November 2015.
Understanding Usages by Modeling Diversity over Time (pdf) (slides)(poster)
Amaury L'Huillier, Sylvain Castagnos and Anne Boyer
In extended proceedings of the 22nd Conference on User Modelling, Adaptation and Personalization (UMAP 2014). Aalborg, Denmark, July 2014. (Late-breaking results session, acceptance rate: 40%)
Utilité et perception de la diversité dans les systèmes de recommandation
Sylvain Castagnos, Armelle Brun and Anne Boyer
10ème COnférence en Recherche d'Information et Applications (CORIA 2013). Neuchâtel, Suisse, Avril 2013. (acceptance rate: 20%)
Eye-Tracking Product Recommenders' Usage (pdf) (slides)
Sylvain Castagnos, Nicolas Jones and Pearl Pu
In proceedings of the 4th ACM Conference on Recommender Systems (RecSys 2010). Barcelona, Spain, September 2010. (acceptance rate: 19.4%)
Recommenders' Influence on Buyers' Decision Process (pdf) (poster)
Sylvain Castagnos, Nicolas Jones and Pearl Pu
In proceedings of the 3rd ACM Conference on Recommender Systems (RecSys 2009), New-York City, NY, USA, October 2009. (acceptance rate: 34%)
Critiquing Recommenders for Public Taste Products (pdf) (poster)
Pearl Pu, Maoan Zhou and Sylvain Castagnos
In proceedings of the 3rd ACM Conference on Recommender Systems (RecSys 2009), New-York City, NY, USA, October 2009. (acceptance rate: 34%)
Probabilistic Reinforcement Rules for Item-Based Recommender Systems (pdf)
Sylvain Castagnos, Armelle Brun and Anne Boyer
In proceedings of the 18th European Conference on Artificial Intelligence (ECAI 2008), Patras, Greece, July 2008
Probabilistic Association Rules for Item-Based Recommender Systems (pdf)
Sylvain Castagnos, Armelle Brun and Anne Boyer
In proceedings of the 4th European Starting AI Researcher Symposium (STAIRS 2008), in conjunction with the 18th European Conference on Artificial Intelligence (ECAI 2008), Patras, Greece, July 2008
(Long version of the paper accepted to ECAI 2008)
Modeling Preferences in a Distributed Recommender System (pdf)
Sylvain Castagnos and Anne Boyer
In proceedings of the 11th International Conference on User Modeling (UM 2007), Corfu, Greece, June 2007
Personalized Communities in a Distributed Recommender System (pdf)
Sylvain Castagnos and Anne Boyer
In proceedings of the 29th European Conference on Information Retrieval (ECIR 2007), Rome, Italy, April 2007
A Client/Server User-Based Collaborative Filtering Algorithm: Model and Implementation (pdf)
Sylvain Castagnos and Anne Boyer
In proceedings of the 17th European Conference on Artificial Intelligence (ECAI 2006), in the 4th Prestigious Applications of Intelligent Systems special section (PAIS), Riva del Garda, Italy, August 2006
(Nominated for the best paper award)
From Implicit to Explicit Data: A Way to Enhance Privacy
Sylvain Castagnos and Anne Boyer
In Workshop on Privacy-Enhanced Personalization (PEP 2006), in Conjunction with the Conference for Human-Computer Interaction (CHI 2006), Montréal, Canada, April 2006
Vers un filtrage collaboratif distribué : le modèle RSB (pdf)
Sylvain Castagnos, Anne Boyer and François Charpillet
3èmes journées francophones sur les Modèles Formels de l'Interaction (MFI 2005), Caen, France, Mai 2005

Other Peer-Reviewed Conferences and Workshops

Don't burst blindly: for a better use of natural language processing to fight opinion bubbles in news recommendations(pdf)
Evan Dufraisse, Celina Treuillier, Armelle Brun, Julien Tourille, Sylvain Castagnos and Adrian Popescu
PoliticalNLP 2022 - First Workshop on Natural Language Processing for Political sciences. Marseille, France, June 2022
Inferring Art Preferences from Gaze Exploration in a Museum
Sylvain Castagnos, Florian Marchal, Alexandre Bertrand, Morgane Colle and Djalila Mahmoudi
In Proc. of the 10th Workshop on Personalized Access to Cultural Heritage (PATCH2019), in conjunction with the 27th ACM Conference on User Modelling, Adaptation and Personalization UMAP2019. Larnaca, Cyprus, June 2019
From Music to Museum: Applications of Multi-Objective Ant Colony Systems to Real World Problems
Pierre-Edouard Osche, Sylvain Castagnos, Anne Boyer
In Proc. of the 11th Workshop on Adaptive and Learning Agents (ALA 2019), in conjunction with the ACM International Conference on Autonomous Agents and Multiagent Systems AAMAS 2019. Montreal, Canada, May, 2019
Modelling students' effort using behavioral data
Barbara Moissa, Geoffray Bonnin, Sylvain Castagnos and Anne Boyer
In Proc. of the International Workshop on Technology-Enhanced and Evidence-Based Education and Learning (TeeL 2019), in conjunction with the 9th International Learning Analytics and Knowledge Conference LAK2019. Tempe, Arizona, March 2019
A Museum App to Trigger Users' Reflection
K. Kontiza, O. Loboda, L. Deladiennee, S. Castagnos and Y. Naudet
In Proc. of the 2th International Workshop on Mobile Access to Cultural Heritage (MobileCH 2018), in conjunction with the ACM MobileHCI conference. Barcelona, Spain, September 2018
Walk the Line: Toward an Efficient User Model for Recommendations in Museums
Pierre-Edouard Osche, Sylvain Castagnos, Amedeo Napoli and Yannick Naudet
In Proc. of the 11th International Workshop on Semantic and Social Media Adaptation and Personalization (SMAP 2016). Thessaloniki, Greece, October 2016
Modélisation de la mémoire à partir du comportement oculaire (poster)
Sylvain Castagnos, Florian Marchal, Dominique Benmouffek and Anne Boyer
1ère conférence sur les approches oculométriques en psychologie. Aix-en-Provence, France, Nov 2015
Utilité de l'Eye-Tracker dans l'évaluation neuropsychologique diagnostique : étude pilote sur une tâche de flexibilité mentale (pdf)
Manon Demange, Florian Marchal, Anaïck Besozzi, Sylvain Castagnos and Ghassan Watfa
Congrès national des unités de soins, d'évaluation et de prise en charge d'Alzheimer (USPALZ 2014). Issy-les-Moulineaux, France, Décembre 2014
When Diversity Is Needed... But Not Expected!
Sylvain Castagnos, Armelle Brun and Anne Boyer
In the 3rd International Conference on Advances in Information Mining and Management (IMMM 2013). Lisbon, Portugal, November 2013
Consumer Decision Patterns Through Eye Gaze Analysis (pdf)
Sylvain Castagnos and Pearl Pu
International Workshop on Eye Gaze in Intelligent Human Machine Interaction, in conjunction with the 2010 International Conference on Intelligent User Interfaces (IUI 2010). Hong Kong, China, February 2010
A positively directed mutual information measure for collaborative flitering (pdf)
Armelle Brun, Sylvain Castagnos and Anne Boyer
2nd International Conference on Information Systems and Economic Intelligence (SIIE'2009), Hammamet, February 2009. (acceptance rate: 39%)
Modélisation des préférences pour un filtrage collaboratif distribué
Sylvain Castagnos and Anne Boyer
Atelier "Intelligence Artificielle et Web Intelligence" (AFIA 2007), Grenoble, France, Juillet 2007
Adaptive Predictions in a User-Centered Recommender System
Sylvain Castagnos and Anne Boyer
In Proceedings of the 3rd International Conference on Web Information Systems and Technologies (Webist 2007), Barcelona, Spain, March 2007
FRAC+: A Distributed Collaborative Filtering Model for Client/Server Architectures
Sylvain Castagnos and Anne Boyer
In Proceedings of the 2nd Conference on Web Information Systems and Technologies (Webist 2006), Setùbal, Portugal, April 2006
A Distributed Information Filtering: Stakes and Solution for Satellite Broadcasting
Sylvain Castagnos, Anne Boyer and François Charpillet
In Proceedings of the 1st Conference on Web Information Systems and Technologies (Webist 2005), Miami, Florida, USA, May 2005

Other Publications

Modélisation de comportements et apprentissage stochastique non supervisé de stratégies d'interactions sociales au sein de systèmes temps réel de recherche et d'accès à l'information (pdf)
Sylvain Castagnos
Thèse de doctorat en Informatique, Université Nancy 2, France, Novembre 2008
Le filtrage collaboratif : Pistes d'applications dans le domaine bancaire et présentation de la technologie
Anne Boyer, Sylvain Castagnos, Yoann Bertrand-Pierron, Jérémy Anneheim, Jean-Philippe Blanchard
Dossiers de la veille technologique du Crédit Agricole S.A., volume 27, Décembre 2006
Proposal of valuation methods for the filtering algorithms
Sylvain Castagnos, Anne Boyer, Jean-Charles Lamirel
Technical report for ASTRA, August 2005
Selection of profiling, filtering and content analysis techniques
Sylvain Castagnos, Randa Kassab
Technical report for ASTRA, March 2005
Bibliography and state-of-the-art report on Filtering and Profiling techniques
Sylvain Castagnos, Randa Kassab
Technical report for ASTRA, March 2005
Des agents collaboratifs pour un filtrage intelligent
Sylvain Castagnos
Mémoire de DEA, Université Henri Poincaré, Nancy, France, Juin 2004

Teaching

Current courses

Ergonomics (Master 2, Cognitive Sciences)
In charge of the module - Lectures and Tutorials
The goal of this course is to present methodologies, tools and knowledge related to Ergonomics, Human-Computer Interaction, Design Thinking, and User Experience. This includes Human Factors (memory, attention, personality, culture, ...), Decision Theory (mental models, Norman cycle), User Studies (qualitative vs. quantitative, between-group vs. within-subject, A/B tests, usability testing, cognitive reasoning, statistics), Audit and Inspections (heuristics, Scapin and Bastien, Nielsen, ...), User-Centered Design (purpose, analysis of population, persona, need assessment questionnaire, task models and diagrams, scenario, use cases, evaluation), and prototyping (Lofi, Hifi, Axure, Gimp, Photoshop, Illustrator, Inkscape). Students put these notions into practice through large-scale projects.
Technologies for Cognitive Sciences (Master 2, Cognitive Sciences)
In charge of the module - Lectures and Tutorials
This is a practical course on eye-tracking technologies, and how to use them in active and passive modes.
Agent-Oriented Programming (Master 2, Miage Speciality 'Distributed Information Systems')
In charge of the module - Lectures and Tutorials
The agent paradigm and the multi-agent systems allow a significant increase of the capacity to model, conceive and build complex distributed systems. The agent-oriented programming comes from Artificial Intelligence and can be seen as an overlay of the object paradigm. It allows to divide a complex problem into simpler sub-problems. The goal of this course is to introduce the notion of agent, its environment, and the purpose of this approach. We also define the interactions and communication processes between agents, the architectures, languages and protocols, the coordination mechanisms and/or the conflict resolutions. The tutorials rely on the platform JADE and the Java programming language. We use Protégé to build ontologies.
Robotics (Master 1, Cognitive Sciences)
In charge of the module - Lectures and Tutorials
This course introduces students with robotics, use of captors and effectors, real time applications, navigation and behaviours. It provides an application framework through the use of Mindstorms EV3 robots with LeJOS (Java).
Software Engineering (Licence 3, Cognitive Sciences)
In charge of the module - Lectures and Tutorials
This course learns students how to manage a development project and do software programming using tools such as SVN, Ant, Maven, JUnit, design patterns. It also deals with advanced Java programming (Threads, RMI, ...).
Object-Oriented Programming (Licence 2, Cognitive Sciences)
In charge of the module - Lectures and Tutorials
This course outlines the object-oriented programming, and more particularly the Java programming language. These tutorials aim at presenting various notions such as functions and methods, collections, encapsulation, polymorphism, inheritance, interfaces, graphical user interfaces, and database connection. Students have to develop a video game in Java at the end of the module.
Introduction to Ergonomics (Licence 1, Cognitive Sciences)
Lecture
This is an introduction to Ergonomics and Human-Computer Interaction (Kinect, Leap Motion, Eye-tracker, Tablet).
C2i - Office Automation Tools Part 1 and 2 (Licence 1, Campus LSH)
Lectures and Tutorials
Following the module "Office Automation Tools Part 1", this course prepares students to the certificate of Computer Science and the Internet (C2i). We introduce the following notions: the HTML language, image manipulation program (GIMP), presentation program (Microsoft Powerpoint), spreadsheet application (Microsoft Excel) and the collaborative work with a word processing software (Microsoft Word).

Previous courses

Modelling Users' Behaviour and Social Interactions (Master of Research, IPAC Speciality)
In charge of the module - Lectures and Tutorials
Based on my research activities, this course introduces stacks and solutions when managing information and knowledge. It particularly focuses on data mining techniques and recommender systems: Bayesian networks, collaborative filtering, content-based filtering, etc.
Communication and Information Systems (Master 1, LEA)
In charge of the module - Lectures and Tutorials
Students learn how to communicate and interact efficiently on the Internet. Lectures alternate practical (HTML, JavaScript, PHP, configuration and use of the CMS Wordpress, RSS feeds) and theoretical notions (Client/Server Architecture, Web Hosting and Domain Names, Web indexing, functioning of search engines, social networks, workflow, ...).
ABCDWeb (E-Business) (ARTEM, ENSMN 2A, INPL)
Audit of the projects
This course aims at combining interdisciplinary competences of students coming from ENSMN, ICN and ENSA within the frame of real industrial projects. They also learn how to follow management project rules (engagement letter, need assessment, specifications, risk analysis, planning, development, ...).
Advanced algorithmics (ENSMN 2A, INPL)
Teaching notes and Lectures
This course learns students how to easily implement multi-agent systems for different application domains such as patrolling, simulation, information systems, video games, ...
Human Computer Interaction (Master, EPFL)
In charge of the module - Teaching notes, Lectures and Assignments
This course is designed to provide complimentary programming skills to those who have been trained in software engineering. In particular, it gets students started with interaction design and user-centered design (human factor theories, problem space, conceptual models, need assessment questionnaire, task model, task frequency model, action table, priority assigment of tasks, usability goals, constraints, user scenarii, ...).
Database Query Languages (Licence 3, Miage)
In charge of the module
This course bridges the gap between Relational Algebra, Predicate Logic, Predicate Languages (tuples, domains) and SQL. Exercises rely on Access, MySQL and Sybase.
C2i - Office Automation Tools Part 1 (Licence 1, Campus LSH)
Lectures and Tutorials
This module aims at presenting common office automation tools: workspace (ENT University of Lorraine), word processing software (Microsoft Word), operating system (Microsoft Windows), electronic mail, the Internet.
Algorithmic Language for Artificial Intelligence (Master 1, Miage)
Tutorials
An introduction to Artificial Intelligence through concrete case studies: graph theory, expert systems, genetic algorithms, markov models, and so on.
Enterprise Application Components - EJB (Master 2, Miage Speciality 'Distributed Information Systems')
Lectures and Tutorials
This course introduces concepts and procedures allowing a powerful development of enterprise applications. Indeed, in the case of ordinary programming, the upgrade and/or modification of applications are difficult. On the contrary, the constructive programming reduces the needs of technical skills, focuses on the business logic, and makes the application independent from the execution platform. We highlight the benefits of this approach by refering to the Enterprise JavaBeans technology (EJB). EJB components are integrated within a multi-tier architecture (presentation tier, logic tier and data tier).
SQL (Continuing education)
Lectures and Tutorials
This course links the relational algebra with SQL, the latter being the language the most used within database management systems. The exercises rely on Microsoft Access and Sybase.