From: www.itworld.com

KnowledgeMail finds answers in the trash

March 5, 2001 —

E-mail systems highlight a unique problem information-based cultures have: information saturation. It's often the case that more than half the e-missives in a person's inbox are irrelevant to the recipient. Yet at the same time, it cannot be said that more than half the e-mails written are irrelevant. On the surface this may seem paradoxical, but the gap is caused by the problem of distribution. Put simply, getting the right information in the right person's hands at the right time is a difficult task.

As it stands today, e-mail systems contain a wealth of badly distributed information. For example, a message could hold the key for a salesperson to find product experts to be part of a proposal team for an upcoming show, but that salesperson never got it because the sender didn't think that person cared. Ideally, employees would be able to access every tidbit of information that ever passed through your system, whenever they wanted it and regardless of whom it was originally sent to.

We believe Tacit Knowledge Systems has come up with a very good solution with KnowledgeMaiil 2.0. This fascinating product mines e-mail messages and electronic documents for key phrases and concepts and then builds a searchable database. Users can learn which colleagues have similar interests or information required for specific projects. Additionally, the same database assists help desk agents and others target e-mail requests to those most likely to hold answers.

Further, KnowledgeMail would be a capable complement to Plumtree Software's Plumtree Corporate Portal and Lotus K-Station (with plug-ins for both systems) because knowledge management portals do a fine job of amassing information from other types of databases, but typically not from e-mail.

Identifying expertise

Setting up our KnowledgeMail test system comprising 100 users wasn't difficult and required less than two hours. After loading the server software and connecting to a Microsoft SQL 7 database, we registered users either by hand using a Web form or by automatically drawing information from our Microsoft Exchange 5.5 server (KnowledgeMail also works with other products that support the LDAP directory standard, including Lotus Notes on NT, Windows 2000 Advanced Server, and Exchange 2000). Once done, KnowledgeMail automatically started analyzing e-mail and attachments, a bcc (blind carbon copy) of which is sent to the Tacit server and then deleted to protect users' privacy.

The analysis, fundamental to the whole system, profiles each user by creating a list of searchable public and private terms. Say you correspond with colleagues in Canada about developing a French version of their Web site. KnowledgeMail then surmises that you have experience in international Web development and notes that in the database. From time to time, we examined our profile to determined if the data was correct (which it was) and then decided if we wanted to make our experience public; new terms are kept private by default.

To use the database, we simply entered key phrases in the search box on the portal's home page. The search returned people whose profiles contained matching public terms, weighted by how often the term appeared. For example, when we searched for EDI (electronic data integration) and XML, the top results KnowledgeMail returned were for individuals who had designed a Web application for exchanging purchase order data -- important facts the software gleaned from e-mail and white papers these individuals had sent to others.

What's more, the Search By Example function let us enter up to 40K of free-form text as the basis of our search: "I want to find people who have developed an extranet Web site that uses XML to transmit purchase orders," which is often easier than trying to figure out which keywords are appropriate.

Guarding user privacy

As noted earlier, the system's database holds private information, too. Search results sometimes indicated additional matches based on these private profiles but would not reveal the identity of the authors. KnowledgeMail's Contact Them feature let us compose a message to these anonymous users who could then decide for themselves whether they were the correct contact and reveal their identities.

Yet another problem KnowledgeMail helps solve is routing standard e-mail to the right people. For illustration, we received a message from an employee in another division inquiring about who handled Web application development in Australia -- a question that we couldn't answer. Rather than blast off a note to a large distribution list, we created a new message in Microsoft Outlook that simply asked, "Who manages Web services in North Ryde, Australia?" Next, using the KnowledgeMail Search button (which was added to our mail client's toolbar during setup) the software looked for those recipients who were most likely to know about the question we typed and let us automatiically add their names to a very focused distribution list.

This feature uses the same discovery and ranking system as the KnowledgeMail portal. As a result, the e-mails people receive are usually relevant to their interests and expertise.

After automatically capturing e-mail content, KnowledgeMail makes this expertise available organizationwide through multiple user interfaces. This system complements data that might reside in enterprise sales and support applications or the Web, helping users become more responsive to many business situations. This can have a large financial benefit as clients receive better service, and it can turn your e-mail systems into higher-performing assets.

InfoWorld