Center for Automated Learning and Discovery
School of Computer Science, Carnegie Mellon University


Inferring Users' Projects
from their Workstation Contents

Tom M. Mitchell, Sophie H. Wang, Yifen Huang, Adam Cheyer

November 2005


Keywords: Email analysis, Workstation content analysis, Desktop Search, ActivityExtractor, clustering, social network analysis, machine learning

One key to providing intelligent assistance to workstation users is to construct machine-understandable descriptions of the user's ongoing projects, or activities, (e.g., their committee memberships, writing projects, conference organization activities), and indices describing which emails, meetings, and colleagues relate to which activity. This paper presents a program, ActivityExtractor, which examines the user's workstation contents to infer such activity descriptions. In earlier work we described an algorithm which infers activities by clustering the user”Ēs emails based on their word distributions. Here we extend this approach in several ways: (1) by incorporating a social network analysis of email senders and recipients, (2) by considering workstation contents beyond email, including contents accessible via Google Desktop Search, and (3) by allowing simple user input in the form of a list of activity/project names. We describe the ActivityExtractor algorithms and report on experiments applying these to several users' workstations.

21 pages

SCS Technical Report Collection
School of Computer Science homepage

This page maintained by