Collaborative filtering is one of the information filtering techniques, that recommends information referring to the evaluation of others' feedback, where their preferences are similar to the target user to assist. The user's preference is often specified explicitly, and it is often a burden or disturbance in concentrating on his/her primary activity. This issue affects the granularity, i.e., degree of detail of user's feedback, since it largely depends on the easiness of acquisition for the user's preference. In this paper we describe a method of information recommendation based on social filtering, where the preference of user is implicitly acquired by gaze detection. As an example of the application we implemented a system that recommends paintings to a person based on others' attention in appreciating paintings. The evaluation of preference is based not on individual objects (i.e., paintings) but on sub-regions in an object (e.g., a person, a building, an animal, and so on), that is detected by gaze point tracking. The strength of interest is measured as the duration of watching a sub-region in the painting, and each user's interest model is organized based on it. Experimental result showed the sub-region based information recommendation provides us better recommendation compared with object-based recommen-dation, and the proposed implicit preference acquisition method is comparable to explicit preference specification method.
[1]
Tsukasa Hirashima,et al.
Information Filtering Using User's Context on Browsing in Hypertext
,
2004,
User Modeling and User-Adapted Interaction.
[2]
Bradley N. Miller,et al.
GroupLens: applying collaborative filtering to Usenet news
,
1997,
CACM.
[3]
Yiyu Yao.
Measuring retrieval effectiveness based on user preference of documents
,
1995
.
[4]
Thorsten Joachims,et al.
Web Watcher: A Tour Guide for the World Wide Web
,
1997,
IJCAI.
[5]
Katia P. Sycara,et al.
WebMate: a personal agent for browsing and searching
,
1998,
AGENTS '98.
[6]
Bernard Mérialdo,et al.
Using category-based collaborative filtering in the Active WebMuseum
,
2000,
2000 IEEE International Conference on Multimedia and Expo. ICME2000. Proceedings. Latest Advances in the Fast Changing World of Multimedia (Cat. No.00TH8532).
[7]
Yoav Shoham,et al.
Fab: content-based, collaborative recommendation
,
1997,
CACM.
[8]
Douglas B. Terry,et al.
Using collaborative filtering to weave an information tapestry
,
1992,
CACM.
[9]
Susan T. Dumais,et al.
Personalized information delivery: an analysis of information filtering methods
,
1992,
CACM.