Design and implementation of a user modeling framework

From Master Projects
Revision as of 10:08, 8 March 2013 by Mcaklein (talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

has title::Design and implementation of a user modeling framework
status: finished
Master: project within::Cognitive Science
Student name: student name::Patrick Sannes
Start start date:=2010/02/01
End end date:=2010/07/31
Supervisor: Michel Klein
Second reader: has second reader::Tibor Bosse
Company: has company::TNO
Thesis: has thesis::Media:Thesis.pdf
Poster: has poster::Media:Media:Posternaam.pdf

Signature supervisor



Introduction At TNO Human Factors we develop intelligent agents that can support users in a variety of ways. Some agents act as virtual assistants, others provide task support in an (in)visible way, while again others play a role (visualized as non-visualized) in the training of people. What all these agents have in common is that they interact with a human that requires assistance, task support or training. In order to optimize this interaction, it is important that the agent has a sense about the user. In case of virtual assistants the agent has to know about the likes and dislikes of its boss; in case of supportive agents when to give support (e.g., not when the task executor is very busy and should not be distracted); and for training what the learning objectives of the student are, so it knows how to react to optimize training.

Aim Within this M.Sc. project the student will work on the development of a generic framework with which various types of users can be modeled by means of several model dimensions. The framework will probably contain several parts or layers to discriminate between the modeling of 1) the users actual state (busy, stressed, tired…); 2) its personality (likes to be handled kindly and not ordered what to do, likes spaghetti and not broccoli, wants feedback in text or speech, has only one hand, …); and 3) specific knowledge about the users task (when X  Y is correct, the user requires information about the weather in the morning, the user has to learn Z, …). In addition, there has to be a mechanism that can reason about the several dimensions, and integrate them to act towards the user in an optimal way.

In order to develop this framework the student will first perform a literature study on the various ways that are proposed to model users in several domains (from e.g. intelligent tutoring systems and adative support to personalized web browsing). Next, the student will work on the formalisation of a framework that supports and integrates the modeling of a user on multiple dimensions. Last, the student will built a prototype of the framework for a specific domain, in which the several dimensions of the user model are combined to come to an optimal human-agent interaction.