Recommendation Models of Television Program Genre, Based on Survey and Analysis of Behaviour in Watching Television : Toward Human Content Interface Design (2)

• Takahashi Yasushi

  Sony Corporation

• Shackleton John

  Chiba University

Bibliographic Information

Other Title

• テレビ視聴行動の調査分析に基づく番組ジャンル推薦モデル : ヒューマン・コンテンツ・インタフェースデザインに向けて(2)

Abstract

Aimed at developing an automatic program recommendation system to meet individual television viewer's preferences, two genre recommendation models based on age and gender, and on factors of attitude and sense-of-value in television viewing respectively, were developed using nonlinear canonical correlation analysis and an analysis of television viewing ratings with age and gender. The main points found regarding the construction of the recommendation models are as follows ; 1. A system applying the two models simultaneously to the general database appears to produce effective genre recommendation. 2. The general database consists of a data table of three ranked genre recommendations for each age and gender, and one or more genre recommendations for each viewing scenario, based on factors of attitude and sense-of-value in television viewing. 3. The general database and system learning/inference by viewing history augment each other in genre recommendation. 4. The priority of the two models should be selected in accordance with the particular characteristics of the viewing scenario.

Journal

• Bulletin of Japanese Society for the Science of Design

  Bulletin of Japanese Society for the Science of Design 46 (1), 71-80, 1999

  Japanese Society for the Science of Design

Keywords

• Human Interface
• Watching Television
• Automatic Recommendation

Details

• CRID: 1390001205410838784
• NII Article ID: 110003825100
• NII Book ID: AN00150292
• DOI: 10.11247/jssdj.46.71
• ISSN: 21865221
• ISSN: 09108173
• Text Lang: ja
• Data Source:
  • JaLC
  • CiNii Articles
• Abstract License Flag: Disallowed