Deep Email Miner



The Deep Email Miner Application is a software solution for the multistaged analysis of an Email Corpus. Social network analysis and text mining techniques are connected to enable an in depth view into the underlying information.

Written in Java, published under the GNU GPL and hosted by Sourceforge, it runs under Windows, Linux and any other operating system with Java Runtime Environment 5.0 or later installed.


09.06.2006 - Version v1
The release of the Deep Email Miner Application v1 marks the final version of the first development stage. We are looking forward to many more work on this project in the future. Until then, this homepage will provide the runnable application itself, as well as the program documentation.
If you are interested in further developing the application, we would be very pleased to get in touch with you. Either way, the program sources are now available as an archive from the download page.
The new program version integrates some basic statistical text analysis with the open source Java library WVTool. The refactoring of the email cache component and a good deal of different bugfixes highly improve the program's reliability. Last but not least, the new software manual gives a detailed description of the user interface and the underlying functionality.
26.05.2006 - Version 0.7
New features in today's version: A full working thread detection, the labelling user interface, zoom buttons and many small (but nice) features to the user interface, improving the handling of the network and it's emails.
In the meantime, I'm more and more focusing on writing the software manual, the application report and a project presentation.
Last but not least: there are new screenshots on the homepage. Enjoy!
19.05.2006 - Version 0.6
This week's update introduces the first rudimentary text mining features. The email thread detection searches for chains of replying emails on the same subject. The detected threads can be listed and further explored.
As the base for future text mining techniques, the newly implemented labelling user interface supports the user in the classification of training- and test-data. These labels can also be saved to/loaded from a file. In the future, this component will be connected to the thread detection to further improve the labelling process.
Last but not least, the email viewer has been modified to search for all recipients of the same email message and a lot of bugs have been fixed.
12.05.2006 - Version 0.5
Apart from some smaller changes and bugfixes, the new release adds a lot of new features which strongly improve the analysis process.
The new viewers for vertex and edge sets give a short overview over the network's components, whereas the email viewer enables the user to have a deeper look to the actual data. The enhanced clustering mode with cluster grouping and cluster viewer and the added ranking functionalities allow new ways of viewing the network.
In addition to that, it is now possible to configure many program settings via a configuration file.
05.05.2006 - Version 0.4
Today's new version offers a variety of new features. In addition to some changes to the clustering tab, the new filtering tab has been introduced. Edges in the social network can now be filtered depending on the date and recipient type (TO, CC, BCC) of their emails as well as on the number of contained emails.
Furthermore, the database reader has been strongly debugged and is now fully capable of building networks from the database and displaying email sets of all kinds. This can best be seen in the new node/edge context menu.
Last but not least, the new email cache increases the performance by minimizing the number of necessary database queries.
28.04.2006 - Version 0.3
Version 0.3 adds a first version of the email message viewer, improved tab controls and fixes several bugs.
21.04.2006 - Initial software release: Version 0.2
The initial version of the DeepEmailMiner has been released. Version 0.2 features a simple GUI with graph drawing, statistics and clustering algorithms. The graph can be either a sample graph or loaded from a Mysql database.
04.04.2006 - Project hosted by
As of today, the development process data will be hosted by the Open Source software development web site using a publicly available Subversion programming data repository.
30.03.2006 - Development Started
The development of the Deep Email Miner as a project at the UTS started today. The first component, a social network analysis-module, bases on the Java Universal Network/Graph Framework JUNG.