Search   |   Back Issues   |   Author Index   |   Title Index   |   Contents

Conference Report


D-Lib Magazine
September/October 2007

Volume 13 Number 9/10

ISSN 1082-9873

Contextualized Attention Metadata

Personalized Access to Digital Resources


Jehad Najjar, Martin Wolpers and Erik Duval
Katholieke Universiteit Leuven (K.U.Leuven), Belgium
{Jehad.Najjar, Martin.Wolpers, Erik.Duval}

Red Line


Enabling effective and efficient access to relevant digital resources is one of the key challenges in digital libraries. One approach to addressing this challenge is based on Contextualized Attention Metadata (CAM) that describe how people interact with information (what they read, watch, listen to, publish, etc.) in different contexts. CAM streams are collected from server-side sources like digital repositories, and client-side sources like office software and web browsers.

The CAMA 2007 joint IEEE/ACM JCDL workshop focused on the use of contextualized attention metadata for the personalization of access to digital resources. Approximately 20 researchers and practitioners interested in fostering work on attention metadata in different disciplines shared ideas, experiences and research findings. More specifically, several techniques, architectures and services of (contextualized) attention metadata were presented.

In his opening keynote, Seth Goldstein, the Co-Founder and Chairman of, discussed the challenges and opportunities of collecting and managing attention metadata from the commercial perspective. Seth illustrated his arguments with several services and tools that exploit mainly web-based attention metadata. For example, the AttentionTrust Firefox extension (ATX) collects data about web browsing; this data is then sent to an attention repository to (semi-)automatically establish communities based on who spends attention on what. Another example is AttenTV, where people broadcast their click streams. An important message was that the interest in publishing data about our daily attention on the web is increasing; the main motivation is that people want to measure their influence; e.g., by knowing who consults their attention streams. In the academic world, this would be equivalent to knowing who reads and cites which paper. The focus on how to "break into mainstream" with attention-based applications (for instance through facebook widgets) grounded the remainder of the day in the hard realities of sustainability and scalability in the real world.

In the afternoon keynote, Wayne Hodgins strategic futurist and "corporate evangelist" at Autodesk, Inc., discussed visions of the future for contextualized attention metadata. One important message was that we should concentrate on innovative approaches rather than continue to perfect current irrelevant systems. In particular, new approaches need to take into account that each user is unique at each moment. In this context, contextualized attention metadata is a prime enabler of mass-customization as it allows the capturing of rich user interactions.

The first regular paper session of the workshop dealt with Attention Aware systems:

Joona Laukkanen (American University Paris, France) presented Modelling Tasks: A Requirements Analysis Based on Attention Support Services, co-authored with Claudia Roda and Inge Molenaar. The focus of this work, part of the AtGentive project (, is on identifying user context based on user tasks. The aim of the project is to support the management of the attention of young or adult learners in the context of individual and collaborative learning environments, with artificial agents.

Joe Pagano (Library of Congress, USA) presented Measuring Audience Attention across Multiple Channels. This paper analyzes traffic data for a new section of the Library of Congress web site. Apart from determining the attention and target audience of each webpage, the study measured the sources of traffic to the new website. The results revealed that search engines are not always the most important source of traffic to a web site. Other sources like weblogs or specialized websites can sometimes be more important.

The second session focussed on services and tools for using attention metadata:

Elizeu Santos-Neto (University of British Columbia, Canada) presented Tracking User Attention in Collaborative Tagging Communities, co-authored with Matei Ripeanu and Adriana Iamnitchi. This paper analyzes the distribution of tagging activity in two collaborative communities for managing scientific literature (CiteULike and Bibsonomy). The authors define new metrics for similarity in user interest and use. They suggest using the observations about tagging behaviour to simplify content retrieval and navigation.

Martin Memmel (DFKI, Knowledge Management Department & University of Kaiserslautern, Germany) presented Sharing CAM to Support Personalized Information Retrieval, co-authored with Andreas Dengel. The paper describes a service oriented architecture that allows the collection and integration of various user observation components in workplace scenarios.

Eleftheria Tomadaki (Open University, UK) presented Attention Metadata Visualizations: Plotting Attendance and Reuse, co-authored with Peter J. Scott and Kevin A. Quick. This paper describes work in progress on using attention metadata to track user activities in videoconferencing. User participation in videoconferencing events and replays of recorded events is tracked and visualized.

The workshop concluded with a discussion of what participants found relevant for their own work at their home institution. Some of the interesting observations from that discussion include:

  • Blogs are becoming an important source for web traffic.
  • Social communities can be identified through attention metadata, for example, based on the tags people provide and accesses to similar material.
  • There is an emerging business activity around attention metadata.
  • There are some early approaches and schemas for integrating attention metadata collected from various applications. Some approaches consider sources of attention metadata like eye tracking and sensor data.
  • The value of attention metadata derives from its capability of enabling access to the appropriate information at the right time, right place and in the right medium. In addition, attention metadata is an interesting enabler of feedback about activities, e.g., helping people understand who is reading what they publish.
  • Little work is done on processing attention metadata. Most of the current work focuses on collecting, and to some extent, merging this data.
  • Little work was presented on the privacy issues surrounding attention metadata.

The presentations from the CAMA Workshop were recorded and are, along with the papers and slides, accessible at <>. Proceedings of the workshop are published at <>.

Copyright © 2007 Jehad Najjar, Martin Wolpers and Erik Duval

Top | Contents
Search | Author Index | Title Index | Back Issues
Previous Conference Report | Next Conference Report
Home | E-mail the Editor


D-Lib Magazine Access Terms and Conditions