| 摘要 |
An apparatus and method to create demographic user profiles of communication network users. In an embodiment, the communication network is the internet, and demographic profiles are produced for internet users using URL data. A probabilistic model is created for each demographic attribute, with example demographic categories including marital status, gender, age, income, profession, industry, etc. Each probabilistic model is computationally independent of the others, and is derived by generating interest scores for a filtered URL set, identifying a discriminating URL set for each demographic attribute, and creating a probability distribution function from the respective interest scores. The probabilistic models accept an interest score and produce a corresponding demographic probability measure that indicates the probability that the respective models exist for each demographic attribute, user profiles may be maintained and updated by obtaining a specified user's URL data, filtering the data, attaching interest scores to the filtered data, inputting the interest scores to respective demographic models to produce probabilities, and updating the user profile with the newly computed probabilities.
|
