Data and Knowledge Processing (DKPro) - A middleware for natural language processing applications

• Applicant: Professor Dr.-Ing. Torsten Zesch
• Subject Area: General and Comparative Linguistics, Experimental Linguistics, Typology, Non-European Languages ; Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: from 2018 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 390886566

DKPro (Data and Knowledge Processing) is a research software in the field of computational linguistics / natural language processing with a world-wide user base. It bridges the gap between tool providers that develop research prototypes on one side and on the other side researchers who want to use those prototypes as well as developers that seek to integrate NLP technology into their applications. Thus, DKPro plays an important role in the community as aggregator and integrator of research software that cannot be easily replaced, similar to the role that Linux distributions play in the realm of operating systems. Despite being continuously developed since almost 10 years now, the development of DKPro rests mainly on a few individuals. Progress towards a well-defined community process and community governance to eventually provide a community-driven sustainability model is present but very slow. Therefore, this proposal is designed to provide the resources to accelerate and drive this process and generate sufficient momentum over the course of three years that runs self-sustained afterwards. We propose a set of work packages that target both technical and administrative issues with the goal to establish a large user and developer community that will enable a continued development and maintenance after the funding period. One major technical gap, that cannot be addressed with the currently available resources, is the lack of interfaces to big infrastructure projects like CLARIN-D and to user-facing analytics environments like KNIME or RapidMiner. On the administrative side, a main goal is to establish an umbrella organization with a clear community governance concept that will drive the further development and engage the community through workshops and seminars.We believe that DKPro is an excellent candidate for funding under this call, as its small but active developer community has already identified and discussed the problems of sustainable development and are ready to contribute to the proposed work packages. We contribute considerable experience not only at the technical level as authors of the DKPro software, but also on the organizational level or on the level of research and open-source communities, where we have been actively participating and contributing over many years.DKPro already has a highly professional set of workflows for archiving, versioning, issue tracking and documentation as well as a licensing model that is suitable for academia and industry. DKPro is thus a good example for a community-driven transformation process, which we plan to document as a reference example for later similar cases.

• DFG Programme: Research data and software (Scientific Library Services and Information Systems)