Interoperability of corpus processing work-flow engines: the case of AlvisNLP/ML in OpenMinTeD

AlvisNLP/ML is a corpus processing engine developed by the Bibliome group. It has been used in several experiments and end-user applications. We describe its design principles and data and workflow models, then we discuss interoperability challenges in the context of the OpenMinTeD project. The objective of OpenMinTeD (EC/H2020) is to create an infrastructure for Text and Data Mining (TDM) of scientific and scholarly publications. In order to offer to the infrastructure users a single entry point and the widest range of tools as possible, the major European corpus processing engines will be made interoperable, including Argo, DKPro, and GATE. We show that AlvisNLP/ML can be fully integrated into the OpenMinTeD platform while maintaining its originality.