TY - GEN
T1 - A novel web page analysis method for efficient reasoning of user preference
AU - Lee, Seunghwa
AU - Jung, Minchul
AU - Lee, Eunseok
PY - 2008
Y1 - 2008
N2 - The amount of information on the Web is rapidly increasing. Recommender systems can help users selectively filter this information based on their preferences. One way to obtain user preferences is to analyze characteristics of content that is accessed by the user. Unfortunately, web pages may contain elements irrelevant to user interests (e.g., navigation bar, advertisements, and links.). Hence, existing analysis approaches using the TF-IDF method may not be suitable. This paper proposes a novel user preference analysis system that eliminates elements that repeatedly appear in web pages. It extracts user interest keywords in the identified primary content. Also, the system has features that collect the anchor tag, and track the user's search route, in order to identify keywords that are of core interest to the user. This paper compares the proposed system with pure TF-IDF analysis method. The analysis confirms its effectiveness in terms of the accuracy of the analyzed user profiles.
AB - The amount of information on the Web is rapidly increasing. Recommender systems can help users selectively filter this information based on their preferences. One way to obtain user preferences is to analyze characteristics of content that is accessed by the user. Unfortunately, web pages may contain elements irrelevant to user interests (e.g., navigation bar, advertisements, and links.). Hence, existing analysis approaches using the TF-IDF method may not be suitable. This paper proposes a novel user preference analysis system that eliminates elements that repeatedly appear in web pages. It extracts user interest keywords in the identified primary content. Also, the system has features that collect the anchor tag, and track the user's search route, in order to identify keywords that are of core interest to the user. This paper compares the proposed system with pure TF-IDF analysis method. The analysis confirms its effectiveness in terms of the accuracy of the analyzed user profiles.
KW - Recommendation system
KW - TF-IDF
KW - User preference
KW - User profile
UR - https://www.scopus.com/pages/publications/48949102668
U2 - 10.1007/978-3-540-70585-7_10
DO - 10.1007/978-3-540-70585-7_10
M3 - Conference contribution
AN - SCOPUS:48949102668
SN - 3540705848
SN - 9783540705840
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 86
EP - 93
BT - Computer-Human Interaction - 8th Asia-Pacific Conference, APCHI 2008, Proceedings
T2 - 8th Asia-Pacific Conference on Computer-Human Interaction, APCHI 2008
Y2 - 6 July 2008 through 9 July 2008
ER -