Comparing feature-based and clique-based user models for movie selection

Abstract

The huge amount of information available in the currently evolving world wide information infrastructure at any one time can easily overwhelm end-users. One way to address the information explosion is to use an "information filtering agent" which can select information according to the interest and/or need of an end-user. However, at present few such information filtering agents exist. In this study, we evaluate the use of feature-based approaches to user modcling with the purpose of creating a filtering agent for the video-on-demand application. We evaluate several feature and clique-based models for 10 voluntary subjects who provided ratings for the movies. Our preliminary results suggest that feature-based selection can be a useful tool to recommend movies according to the taste of the user and can be as effective as a movie rating expert. We compare our feature-based approach with a clique-based approach, which has advantages whcrc information from other users is available.

Untitled

“…A different type of comparative study was performed by Alspector et al (1997) for the domain of ¢lm recommendations. They compared the performance of a recommender system built under the collaborative approach against that of a system built under the content-based approach.…”

“…modifying its dialogue strategy (Litman and Pan, 2000); to recommend objects a user may be interested in, e.g. news items (Jennings and Higuchi, 1993;Billsus and Pazzani, 1999) or ¢lms (Alspector et al, 1997); and to perform actions on behalf of a user, e.g. pre-sending WWW pages or forwarding email (Macskassy et al, 1999).…”

Untitled

“…All of the empirical evaluation papers in the ¢rst nine years of UMUAI can be classi¢ed into four broad topic areas: ten papers in adaptive-hypermedia/information-¢ltering (Jennings and Higuchi, 1993;Kaplan et al, 1993;Boyle and Encarnacion, 1994;Vassileva, 1996;Mathe¨and Chen, 1996;Raskutti et al, 1997;Newell, 1997;Hirashima et al, 1997;Alspector et al, 1997;Balabanovic¨, 1998), nine in student modeling (Nwana, 1991;London, 1992;Carbonaro et al, 1995;Corbett and Anderson, 1995;Kashihara et al, 1995;Shute, 1995;Webb and Kuzmycz, 1996;Mitrovic et al, 1996;Milne et al, 1996;Sison et al, 1998), nine in plan recognition/mixed-initiative interaction (Calistri-Yeh, 1991;Albrecht et al, 1998;Gmytrasiewicz et al, 1998;Chiu and Webb, 1998;Chu-Carroll and Brown, 1998;Guinn, 1998;Ishizaki et al, 1999;Green and Carberry, 1999;Virvou and du Bulay, 1999), and three in user interfaces/help systems (Tattersall, 1992;Krause et al, 1993;Debevc et al, 1996).…”

“…For a review of machine learning and predictive statistical models see Webb et al (2001), and Zukerman and Albrecht (2001) respectively. Adaptive hypermedia and information ¢ltering build on the practice of empirical evaluation of information retrieval systems through measures developed in library sciences such as recall and precision (e.g., Mathe¨and Chen, 1996; Raskutti et al, 1997) and similarity/ relevance metrics (e.g., Newell, 1997;Hirashima et al, 1997;Alspector et al, 1997;Balabanovic¨, 1998). For a review of adaptive hypermedia, see Brusilovsky (1996Brusilovsky ( , 1998Brusilovsky ( , 2001).…”

Predicting Users’ Requests on the WWW

“…Both types of systems require a prediction model which anticipates a user's preferences, including documents a user may find interesting, or his/her future actions. These models are generally obtained by applying machine learning techniques to identify these preferences or future actions based on the preferences or actions of (1) the users themselves (Davison and Hirsch, 1998,Joachims et al, 1997,Lieberman, 1995, (2) a group of similarly-minded users (Alspector et al, 1997), or (3) the general population (Bestavros, 1996.…”