LifeLong User Modeling Workshop

The 2nd Workshop on Lifelong User Modeling took place in Rome on June, 14th, 2013.

Together with Juday Kay and Bob Kummerfeld (both from the University of Sydney), Frank Hopfgartner and me organized the 2nd Workshop on Lifelong User Modeling, in conjunction with UMAP 2013. The half day workshop started with a keynote given by Dr. Cathal Gurrin (Dublin City University) with the title ‘Experience of a Lifelogger: Tasks and Challenges’. Life logging can support us in different scenarios. In medical settings it can help to live healthier by analyzing daily activity data or to help patients with Alzheimer’s disease to remember certain situations. To do so, we have to log as much personal data as possible – movement data, body data etc. Cathal is wearing a Sensecam since 2006 and thereby created over 12.5 million images (plus GPS logs, some audio files and more). In his talk, Cathal presented the daily challenges of a life logger from a personal point of view, wearing a camera arousing other people’s suspicion and sometimes making personal conversations complicated, and from a research point of view. Such big data, as created by a constant logging of personal data, brings new challenges for IR systems. How do we process the data, how do we find information in this big amount of data? In his talk, Cahtal told an anecdote about a small car accident he had and what he remembered about it. In his memory, the chronology of what happened was clear. After the crash he went to the garage, talked with the mechanic, got a rental and went to the university. To make a long story short, the pictures revealed that the chronological order that he memorized was messed up, between the different events there was a week in-between. Cathal also presented some projects where he and his team tried to allow users to interact with the collected data. My favorite project is is the ‘Colour of Life’ wall. The Colour of Life wall is a touchscreen visualization that plots a two dimensional view of a person’s life experience, in terms of colours encountered, on a large video display wall.

After the keynote, we had three full paper presentations:

Johnson Iyilade and Julita Vassileva – A Decentralized Architecture for Sharing and Reusing Life-Logs
Márius Šajgalík et al. – Efficient Representation of the Lifelong Web Browsing User Characteristics
Lie Ming Tang and Judy Kay – Lifelong User Modeling and Meta-cognitive Scaffolding: Support Self Monitoring of Long-Term Goals

Johnson Iyilade presented a decentralized approach for multi-application life-logs sharing and reuse. The idea is to have a decentralized framework which allows applications to share data over a common and secure API. As far as I understood, parts of the framework are already implemented, so we hopefully see more in the near future.
Márius Šajgalík presented a browser based life logging system with captures a users’ interest log based on his browsing behavior. The system is very efficient in ways of storing and accessing the data using radix trees.
Judy Kay, presenting the work of her student, talked about requirements for lifelong user models for self monitoring purposes. Without repeating everything, requirements can be split into three parts:

Capturing and managing the data with focus on data coherence.
Personal data access and usage – Control over the data.
Lifelong usage – How to set long term goals and maintain motivation.

All papers can be found here – CEUR Volume 997

The presented papers show that the topic lifelong user modeling is a broad one. To really have systems that can support users over a life span, questions about how to capture data, how to store and manage it, and how to make it accessible have to be answered. We want to thank all participants for the valuable discussion and the speakers for their interesting talks and hope to meet you all again soon.