Kadmos the graph-based text research system
Kadmos is a scalable backend and experimental platform for developing novel forms of algorithmic interaction with text collections in the Digital Humanities. It is to be made available as free and open source software.
The software has been pre-released in early 2017 and is now undergoing a further development phase until an official release accompanied by documentation material will be available. More details will be given on this page later in 2017.
Application Scenarios
In most text mining and corpus analysis tools the functionality provided out-of-the box is likely not sufficient for a specific use case. New functionality has to be added and constantly refined. Kadmos is no exception in that respect. What makes this backend system special is that it is built to render the adoption process process as easy as possible while providing the user with a flexible data model and access pattern to crate concise interactions with the data.
Download, Installation and Setup
The Kadmos package can be set up on any system running a recent version of JRuby. In a typical Debian-based server environment only a few commands are needed to install the software with all required libraries and start the system:
> apt-get install git jruby > jruby -S gem install bundler > git clone git://git.textgraph.science/kadmos > cd kadmos > jruby -S bundle install > jruby -J-Xmx4g -J-Dfile.encoding=UTF-8 lib/kadmos.rb
The program code can be obtained via git
as shown above, or directly downloaded as a zip file here.
Documentation
Proper end-user and developer documentation will follow with the main release.
Plugins
Several plugins are currently under development:
- A component to make use of the MATE-Tagger along with its language modules for POS tagging and dependency parsing:
- An OAI-PMH interface to publish the metadata of all the documents contained in the imported coprpora
- A native GUI example that demonstrates how to use the Kadmos functionality in a standalone desktop application
Related Publications
- [Efer2015]
- Thomas Efer: Text Mining with Graph Databases - Traversal of Persisted Token-Level Representations for Flexible On-Demand Processing. In: Unger, Herwig; Halang, Wolfgang A. (Eds.): Autonomous Systems - Proceedings of the 8th GI Conference, VDI Verlag, 2015 (Fortschritt-Berichte VDI, Reihe 10 842), S. 157–167 – Download
License
This Kadmos pre-release has not yet been given a well-established open-source license. In the meantime you are free to view, run and modify the code but are not allowed to distribute your copy to a third party without asking for permission.
Acknowledgements
Parts of Kadmos have been created within the BMBF-funded project eXChange and the development continues under the umbrella of the BMBF-funded project ScaDS