ActiveSearch takes a collection of emails (or any dataset where a similarity can be generated between elements) and recommends related messages based on user feedback. The user provides an initial seed email then enters into a cycle where ActiveSearch provides a similar email and the user reports whether or not the email was interesting.

ActiveSearch is useful for anyone navigating a large set of emails and looking for related messages on a specific topic. As it considers the similarities between emails as well as user feedback, it is an improvement in accuracy, time, and effort over basic text search or a brute force search.

Required:

• Linux (the tool is written using cross-platform tools but is only tested on Linux)
• Mysql (the emails are stored in a database)
• REST (used to interact with the tool)

Optional:

• Cytoscape (an optional UI we provide that speaks the tool's REST API)
• Perl (if importing different data into the database)
• Java (for changing the similarity comparison algorithm)

This code repository contains two parts.

The Daemon handles all of the computation and exports a REST interface.

The Cytoscape App is an optional plugin for Cytoscape that implements a graphical front end to the Daemon.