2007年度第T期 未踏ソフトウェア創造事業  採択案件概要



  David J. Farber PM (Distinguished Career Professor of Computer Science and Public Policy, Carnegie Mellon University)











  Search People on the Web with Privacy Preserving




Abstract. The World Wide Web is a simple yet effective medium for users to search, browse,

   and retrieve information. In recent years, there has been a prevalence of search engines that

   can find useful information in the Web. Unfortunately, current search engines cannot search

   and summarize the most important object that may be searched ¡people, because these

   search engines always treat the name of a person as a normal string and thus, lacks a natural

   associated relation between the keyword representing the name and a human being. Moreover,

   they only provide Web pages in ranked order instead of summaries,

   which may consume too much of users time to extract out individual information manually.

   Searching for a person is a very challenging task because any person has his/her own attributes

    and may be difficult to be modeled, especially while searching from a huge data source, i.e., the

   Web. However, once the issue is tackled the award is substantial, not only could we search and

   know people on the fly, but most importantly, the individual information can be applied and may

   largely improve the effectiveness and efficiency of personal oriented systems, such as

   usercomputer interaction, mobile services, social networks, online dating, employee finding,

   e-stores, education and training, and so forth.

   Although there are many existing systems that are designed to be personalized,the individual

   information is only used for the individuals own framework and cannot shared to others.

   Moreover, as far as this researcher knows, there are no existing systems aimed at summarizing

    individual information and model users from the whole Web, mainly due to the difficulty of mining

     such a huge and dynamic database. I propose to develop a unified system which provides a

     convenient interface based on ontology user model (to embrace the recent trend of Semantic

     Web), as a Web service or middleware for other personal oriented systems.

     Furthermore, it should be mentioned that users are uncomfortable in exposing their private

     preferential information to others. Thus, my proposal addresses another challenge issue that

     while developing an effective people search system, usersprivacy should be preserved carefully

     by using effective mechanisms.


The proposal by Zhenglu Yang outlines what could be a major step in web searching for people and their works. The ideas are well  supported and the author has an impressive, at least in number, set of publications. If he can actually achieve what he claims to be able to do, it will be a potential basis for commercialization. The deliverables description is rather brief and should receive a bit  more specificity prior to funding.


I would be happy to take this type of effort on and would have a reasonable belief I could help him achieve the stated aims


This proposed project is of strong interest to me for a number of reasons. They include working on an earlier project in the same area that attempted to use cell phones and information about individual's data and activities. That project failed to become commercially viable for a number of reasons. The proposer is interested in developing a system that is more feasible today due to the advent of ubiquitous wireless capabilities via WiFi and increased penetration of cellular data, I would like to help him succeed.


Also, he recognizes the important of privacy in this area and is willing

 to address this critical problems directly.  It can be difficult from

young engineers to look outside their discipline and consider problems

from other fields, but Yang seems to have recognized the important

 of policy issues to his technical work.





  Copyright(c) Information-technology Promotion Agency, Japan. All rights reserved 2004