Link This | Email this | Blog This | Comments (1)

Saving e-mail through automation
June 5, 2007

Despite my claim that e-mail is out of control (see my May Editor’s Note and earlier blog post), there is hope. That’s the contention of the authors of “Managing E-Mail Overload: Solutions and Future Challenges,” writing in the February issue of IEEE Computer, which I have just found the time to read.

Here is their claim: “With proper application of automatic filtering, clustering, and new user interface metaphors, e-mail can once again become an effective knowledge management tool rather than a source of information overload.”

As for why it has reached it’s present deplorable condition, they write, “We arguably have only ourselves to blame…as our use of e-mail as a multipurpose information-processing system stretches this application far beyond its original intent.”

To what inappropriate uses do we as users put e-mail? According to the authors, there are two key ones: first, we use it as an archival tool, and second, we use it to synchronize real-time communication. Elaborating on this second point, the authors explain that e-mail, like traditional mail, is designed to accumulate messages in a repository for asynchronous processing.

Well, can we all just become smarter e-mail users? The authors don’t think so: “The ubiquity of e-mail and its convenience as a knowledge management tool make it unlikely that users’ behavior will change.”

The authors cite several potential solutions, including rule-based systems such as those embedded in Elm Filter and Procmail. Such rule-based schemes generally require some programming expertise on the part of users to set up rules, but the authors comment that a prototype system called Bifrost lets users define filtering rules with a graphical interface.

The authors present a helpful table of 17 e-mail systems that support manual or fully automated message classification.

The authors note that all these approaches depend ultimately on user creation of a folder structure, which users must often revise. In addition, folder-based schemes often do no more than replace message overload with folder overload. Further, complex messages dealing with multiple topics defy easy assignment to specific folders.

What’s to be done? The authors propose their “Automatic Clustering E-Mail Management System,” which “combines text clustering with a simple hierarchical visual interface that reflects commonalities among messages. The ACEMS-generated folder structure derives directly from messages’ textual content….”

They’ve got data to back up their claims for ACEMS: “In a controlled lab experiment, subjects presented with the system’s clustered e-mail folder structure scored 41% higher on a message-retrieval task than those given a simple ordered list.”

*David Schuff, Temple University; Ozgur Turetken, Ryerson University; John D'Arcy, Towson University; and David Croson, Southern Methodist University.

Posted by Rick Nelson on June 5, 2007 | Comments (1)