Motivation

The ever-growing and dynamic nature of user-generated data in online systems poses obvious challenges on how we learn from such data. The underlying problem is how to adapt, in real time, to multiple simultaneous changes involving individual users, user contexts and the system as a whole. Many algorithms are able to adjust their output to some of these changes in real time, however, this requires that the model has been previously trained on data with very similar phenomena. To adapt to new trends, preferences and other unpredictable phenomena, algorithms must be able to update the underlying model itself, which should preferably happen online, incrementally and in real time. This motivates the research on adaptive methods able to maintain and evolve predictive models over time. Incremental learning algorithms and data stream mining have gained maturity in recent years, however this body of knowledge has not transitioned to predictive user modeling, and although the potential to solve relevant problems is high, advances in this direction are far from trivial. This calls for further investigation and scientific contributions.

Topics

  • Incremental user modeling over multidimensional data streams
  • Incremental learning algorithms for recommender systems
  • User preference change detection and adaptation
  • Context change detection and online adaptation
  • Cold-start in incremental recommender systems
  • Session-based incremental recommender systems
  • Long-term incremental user modeling
  • Incremental modeling with human-in-the-loop
  • Privacy-preserving incremental learning algorithms
  • Explainability of dynamic, evolving models
  • Continuous learning from dynamic content (multimedia, social media, news)
  • Multi-armed bandit algorithms for recommendation
  • Time-sensitive incremental learning
  • Incremental forgetting mechanisms (decremental learning)
  • Online self-tuning algorithms
  • Architectures for continuous user feedback data processing
  • Online evaluation and comparison of incremental algorithms
  • Reproducibility in online incremental methods
  • Reinforcement Learning for dynamic user modeling

Schedule

2020-10-30: Extended abstract submission
2020-11-13: Extended abstract notification
2021-02-05: Full paper submission
2021-04-16: Full paper notification
2021-06-11: Author revisions
2021-07-30: Final notification
2021-09-10: Camera-ready version

All deadlines are at 11:59pm AoE.

Submission

This special issue follows a three stage process. In the first stage, authors are invited to submit an extended abstract of up to 3 pages in the Special Issue submission site. This will be followed by quick review round by the Guest Editors. This abstract must be formatted according to UMUAI's template and guidelines. Authors of accepted papers will then be invited to submit full papers in UMUAI's submission system. A two-round peer review process will then follow. Accepted papers will be published in the Special Issue.

Previously published material

Authors are invited to submit extended versions of their previously published material in conferences and workshops, as well as compilation of such work, provided that they mainly cover one or more topics listed above.

In any case, submissions must follow UMUAI's policy on previously published material. Authors must explicitly refer in the Extended Abstract submission that they are extending previous work and indicate the added value of their submission with respect to the work being extended.

Guest Editors

João Vinagre

LIAAD - INESC TEC
FCUP - University of Porto
Portugal

Alípio Jorge

LIAAD - INESC TEC
FCUP - University of Porto
Portugal

Marie Al-Ghossein

LTCI, Télécom ParisTech
France

Albert Bifet

LTCI, Télécom ParisTech
France

Paolo Cremonesi

Politecnico di Milano
Italy

Contact:

umuai-drsum@googlegroups.com