E-mail patterns map corporate structure

Figuring out a company's power and communication structure may be as simple as examining patterns of e-mail exchanges, according to new research by some Hewlett-Packard scientists.

In the study, researchers attempted

		Reader Resources E-mail administration Tech Republic

to identify different formal and informal communities within an organization by graphing mail flow. Researchers Joshua Tyler, Dennis Wilkinson and Bernardo Huberman studied e-mails sent between any two of the 485 workers at Hewlett-Packard labs over a two-month period, examining 185,773 relevant e-mails in the process.

The researchers said graphing e-mail flow not only correctly identified communities within the organization, but it also provided insight into who the leaders of those groups were. It also helped to identify informal communities that arise when people need to communicate across departments or work collaboratively on projects. What's more, it took just a few hours to analyze the data and identify the groups and their leaders, the study said.

"The power of this method for identifying communities and leadership is in its automation," the researchers wrote. "We have found that it does an effective job of uncovering communities of practice with nothing more than e-mail log ("to:" and "from:") data."

Because it can be captured and stored, many scientists are eyeing e-mail as a tool to quantify exchanges that in the past have taken place in hallways or meetings. The researchers in this study said e-mail flow could provide a window into the communications structure of an organization.

"Given its ubiquity, it is a promising resource for tapping into the dynamics of information within organizations and for extracting the hidden patterns of collaboration and leadership that are at the heart of communities of practice," the study said.

On a practical level, managers, for example, might use information gleaned from e-mail studies to help businesses run more smoothly by making sure teams are communicating effectively and determining who is collaborating on certain projects.

The HP researchers examined the headers of the e-mails they studied to determine who was corresponding with whom. They then used an algorithm that partitions the graph into discrete communities, progressively weeding out correspondence across communities, in order to identify separate groups.

They then interviewed individuals who worked in the lab, asking them to compare the computerized results with the actual department and team structure of the organization. The researchers said that all 16 people interviewed about the process, for the most part, affirmed that the communities identified through e-mail exchanges reflected reality, including groups that weren't officially part of the corporate structure.

The researchers acknowledge that the application of the finding could raise privacy concerns, but they note that the subject of the e-mail is not necessary in order to identify the power structure of an organization.