リヤン マッシュー,西本一志
ユーザの興味の獲得
Abstract:One goal of our work group is to develop a virtual agent, which participates in meetings to supports creative concept formation of the real participants.
A number of human creativity supporting systems acquire user interest in the first place and then provide relevant or partially relevant information according to the user interest in order to support human creativity process. A common way to acquire user interest is by entering a few keywords. Better and more generic ways to acquire user interest is needed to improve human creativity systems.
In this report, two approaches to acquire user interest are proposed, namely a keyword vector approach and a neural network approach. The former approach assumes both the user interest and the available information can be represented by sets of keywords. The latter approach assumes the user interest can be acquired in the weights of a neural network and available information can be represented by sets of keywords.
The experimental results of the two approaches are positive. The results show they probably able to acquire the user interest and maybe even user implicit interest.