UBIQUITOUS KNOWLEDGE DISCOVERY FOR USER MODELING (K-DUUM'07)

Workshop description: Knowledge discovery and data mining are of increasing importance for user modelling. Due to the increasing volumes of data available and the need for satisfactory user adaptivity, knowledge discovery becomes almost a necessity in user modelling. This holds especially in the rapidly growing area ubiquitous computing. By "ubiquity", we mean ubiquitous devices, ubiquitous data (with spatial components), and ubiquitous (distributed) processing. In addition, we also refer to the new heterogeneities in the environments and other context properties of usage, the increasing diversity of global audiences (people ubiquity), and to the goal of information ubiquity, the effective findability and accessibility of information. Among the challenges of integrating these aspects into user-model representation and processing, privacy and security play a central role.

This workshop is part of the activities of the HCI and Cognitive Modelling working group of the EU 6FP project KDubiq. With this workshop, we want to work towards a reference system of ubiquitous knowledge discovery for users and user modelling. We start from Czerwinski et al.'s two-dimensional reference system of "digital memories" (CACM 49(1), 2006). The authors propose to localize application classes with respect to two dimensions: By whom is the information recorded and the application controlled (self or others), and by whom is the information used (self or others). For example, a diary is a self/self application, a tutor is an others/self application, and an obituary is an others/others application. We propose to add a third dimension: by whom is the information processed (people or machines). The aim of introducing this third dimension is to investigate how and to what extent (semi-automatic or fully automatic) KD can contribute to better applications.

As one outcome of the workshop, we propose to describe this enhanced reference system and fill it with application classes, adding to and refining Czerwinski et al.'s classification. We expect insights from (a) application classes that change with the media used, and (b) from application classes whose profile becomes clearer by the inclusion of the third dimension. For example, (a) diary blogs are personal diaries, but they are meant to be, and are, heavily used by others; and (b) augmented memories differ in frequency of usage and assessment of usability depending on whether they report retrieved facts (prices) or inferences (collaborative-filtering recommendations) (Plate et al., Proc. of Adaptive Hypermedia 2006). This model will be tested also in the shared session with EDM.

We invite papers covering aspects of all phases of knowledge discovery that are relevant for KD-based ubiquitous user modelling. Topics of interest include, but are not limited to:

• Automated / machine user modelling
• Collection and formats of content, structure, and usage data in ubiquitous environments
• Mining of content, structure, and usage data in ubiquitous environments
• Semantics of ubiquity and ubiquitous knowledge, including Semantic Web approaches and ontologies
• Memory models, in particular augmented memories and shared memories
• User, context, and action models
• Web Communities
• Ubiquitous knowledge discovery: ethics, cultural heritage, languages, diversity and accessibility
• Personalization for ubiquitous computing
• User control of ubiquitous data
• Explicit and implicit user preferences
• Evaluation of user models and user adaptivity: Methodologies and results
•  Privacy and security
• Ownership and control of ubiquitous data and its challenges of UKD
• Privacy preservation techniques
• Multi-user interaction; support of collaboration; group modelling
• Trust negotiation
• Technology acceptance
• Systems and applications using or needing automated user modelling
• Learning and education, knowledge management (see “shared session with EDM’06” above)
• Medical applications
• E-Commerce
• Domestic Environments and Health Care
• Any other novel or emerging application open for an interdisciplinary approach (UM/KD)

Important Dates

• Paper submissions deadline: February 7, 2007
• Notification: March 14, 2007
• Early registration deadline UM2007: March 19, 2007
• Workshop: June 25, 2007

Proceedings and Programme

The proceedings are available here, and the programme is available here.

Queries

Please address general questions to the DM@UM07 workshop organisers at dm.um07@googlemail.com

Workshop chairs

Program committee

We thank our sponsor