Personal Information Filter

1. Introduction

We can't find our favorite services or necessary information from a considerable number of information sources easily. We have to input at least one key word when we search something. And the result does not always satisfy us. It is very important for us to choose our favorite services and necessary information easier. We have been developing Personal Information Filter (PIF) to solve this problem.

2. Personal Information Filter (PIF)

The final goal of PIF is to choose candidates of favorite services and necessary information for user without his/her explicit requirement. It gives user such services and information when he/she turn on his/her terminal. User can see only selected candidates of services and information and unnecessary services or information don't appear at that time.

3. User Profile and Usage Data

We employ user profile(*) and his/her usage data to make and improve PIF. User profile consists of data which is not changed or is not changed for long time (e.g. birthday, name, address, occupation). Usage data is derived from access log.

At first PIF depends on mostly user profile, and users who have similar profile have similar PIFs. But, usage data improves it, and every user eventually has custom-made PIF without his/her explicit requirements.

User may change his/her favorite. It should appear in usage data. And, what is necessary for user also may change. It should appear in profile or usage data. Therefore PIF should keep on updating itself by profile and usage data.

4. Conclusion

PIF is useful for interactive information services. It should be important key element of agents.

(*) This "profile" may be a little different from the profile in usual information filter.