Changelog • Documentation
This release contains improvements for document processing.
- Improves recognizing correspondents/concerned entities for new documents. It makes now better use of the complete data (address book) when analyzing documents.
- Adds text classification feature based on the stanford-nlp classifier: Docspell can now learn from your existing tagged documents using machine learning algorithms (by looking at the extracted text). For new documents, a tag of a given category is suggested. The learning task can be scheduled periodically. You need to enable it in the Collective Settings page.
- Adds support for French in text extraction and text analysis. The joex artifact size increases by about 100M, because of the included model files.
See the Changelog for more information.