To provide a technique for determining whether or not a qualitatively different group appears when analyzing utilization behavior of users to extract features, separating the users into higher and lower two groups when the different group appears, and supporting the generation of a nonlinear function which characterizes the groups.
A utilization record of the users relating to products, systems, or services provided for a market is stored (212), distribution of usage of content or information pages by the users is totalized to calculate an index value, a chart, or numerical data related to a qualitative or structural feature relating to the interrelationship of utilization behavior of the users (213-222), and the obtained index value, chart, or numerical data is output (223). As a result, when separating the users into a user group belonging to a higher layer and a user group belonging to a lower layer in the meaning of the trend of a utilization frequency, the separation is detected according to the qualitative feature, the boundary between the higher layer group and the lower layer group is defined, and the determination on the occurrence of the separation and the evaluation of the boundary is supported.
COPYRIGHT: (C)2010,JPO&INPIT
JP2003216776A | ||||
JP2003331207A | ||||
JP2002014964A |
市川 裕介,Web行動履歴を用いたユーザ嗜好特性推定技術の検討,マルチメディア,分散,協調とモバイル(DICOMO2008)シンポジウム論文集,日本,社団法人情報処理学会,2008年 7月 2日,第2008巻 第1号,pp.1786-1792
中村 美穂,アクセスログを用いたユーザの革新性推定方式に関する一考察,第83回 知識ベースシステム研究会資料 (SIG-KBS-A802),日本,社団法人人工知能学会,2008年11月14日,pp.27-32