EviDENce Framework processing and model construction

EviDENce Framework processing and model construction

The released docker framework comprises as system to ingest and process a user supplied text corpus, construct a doc2vec model over the elements of the processed corpus enabling abstract content similarity queries, and exposes the ingested, processed corpus and the constructed model for incorporation into the (separate) evidence-gui for further user interaction.

Executing the run_evidence_framework.sh script with input configuration files as specified in the instructions starts a Jupyter lab server with notebooks for corpus ingestion and model construction. The framework currently supports the use of a doc2vec model, but the user can add/substitute other vector space based models and continue to use the framework.

The released framework constitutes a beta test version and, while it has been used in specific applications analysing large corpora of historical dutch text, the user is advised to safeguard and monitor applicability for their purposes.