 |
 |
 |
 |
| |
|||||
|
|
 |
With the volume of Web-based information increasing at an unprecedented pace, information stewards desperately need to be able to evaluate quickly and accurately the usefulness of their Web collections. We are only now beginning to address the question of how to appraise the value of a corporation's on-line knowledge. One important aspect of appraising value is examining the extent to which collections are being used and determining what sorts of employees are using them. Raw hit counts of Web page usage deliver too much information at too high a level of detail. And while a survey completed by every employee would provide a solution to this problem, this method would consume an unacceptable amount of resources. To address this problem, MITRE is developing a tool called Intranet Usage Profiler (IUP) that exploits data already collected in corporate Web logs of intranet usage. IUP underscores an axiom of knowledge management (KM) that an organization's information processing directly correlates with the capacity of its work force to produce innovation. A research company that values creativity like MITRE, is an ideal environment in which to deploy a KM system based on this belief. The advantage of using a KM tool like IUP is that, with IUP, information stewards do not need to require employees to monitor the use of their own collections. Necessary information is extracted from existing Web logs, so the overhead, both financial and computational, is minimal.
IUP provides collection stewards, who are responsible for maintaining sets of MITRE's intranet Web pages, with a tool for profiling usage of Web collections. It is a Web-based application that captures information about the consumption trends of the MITRE community through analysis of Web usage. The contents of the harvested Web logs record the name of the requesting machine, the requested file name, and a time stamp. IUP works by associating who owns a computer, determined from MITRE's property database, with demographic information about the owner obtained through the company's Lightweight Directory Access Protocol "white pages" directory. By combining these results in a database, we can more fully and accurately answer the question "Who is looking at my collection?" For example, IUP can answer all of these questions:
Since the developers of IUP were trying to create an open portal into a collection of private corporate information, IUP is designed to uphold its own self-imposed privacy policy, whose rules are intended to protect the privacy of individual users of the intranet without compromising IUP's viability. Our rules for privacy are twofold: "What you can see is identical to what I can see" and "Nobody can monitor individual usage." IUP implements the first privacy rule of equal views for all users by delivering the information to the user via a Java applet imbedded within its own Web page. Because it is on-line, MITRE employees can access the system through a Web browser and see for themselves the nature of the information that the tool generates. IUP enforces the second privacy rule, ensuring that IUP isn't used as a monitoring tool, by filtering findings from any query when theresult is fewer than three individuals. This means if the number of employees from a division viewing a collection is, for example, two, IUP will not return any information. This type of filtering is designed to prevent a user from cross-referencing multiple queries from the system to find more information than the privacy policy permits. While creating an application that follows these rules limits IUP's functionality, this is an intentional choice made to avoid the possibility that IUP could be used as a monitoring tool. Any slight loss in functionality is more than offset by having a tool that can be used and positively accepted by the MITRE community. Presently, IUP logs demographics of employees viewing pages within the corporate intranet. What we would like to be able to do is identify unique pages that are "authorities" and "hubs" from within these corporate Web collections. "Authorities" are Web pages to which other sites link frequently in order to provide access to information on a particular topic. "Hubs" are sites that cite many of those authorities. IBM has produced some promising new algorithms for doing just this in the new Clever search engine. This work focuses on analysis of hyperlinks within the Web pages. Hyperlinks are usually the blue underlined text inside a Web page that, when "clicked on" by the mouse, send the Web browser to a new page. By recording and analyzing these links between pages, Clever can assign values to different pages. These values can increase a page's rating as either an "authority" or a "hub." If the methods from the Clever project were combined with our own profiling techniques, IUP could rate how proficient the MITRE community is at finding the knowledge it is looking for. This would allow IUP to answer not only "Who is looking at my collection?" but also "How are people looking at my collection?" In addition to these technical issues, the development team would like to address more social concerns. Does implementing a public, company-wide metric system such as IUP have any negative effects? Does deploying a tool such as IUP enhance a company's ability to learn where the on-line information is? By identifying strengths and weaknesses in its corporate knowledge base, does a company truly become more innovative? Disseminating the answers to these and similar questions could make a useful contribution to the fields of knowledge management and information stewardship. Related Information For more information, please contact Robert McCready using the employee directory. |
Solutions That Make a Difference.® |
|
|
