Some scientists & librarians had documents (f.e. "Sammelband") / digitized books for which named entity recognition failed because of default limit (one million chars) of NER lib. Thanks to https://t.co/Kdg5fyM7Tv from Tilburg University for new config option to extend the limit!
Imagine having 1000 PDFs and needing to find those with specific keywords. Here’s a tool many, many journalists need that someone could easily write and share. 🔍🗞
- Take PDFs as input
- Convert to text
- Search for keywords (UTF-8)
- Output result as CSV
Heute gaben @acka47 und @fsteeg im Rahmen eines Kolloquiums der Professur Wirtschafts- und Sozialgeschichte an der @UniHalle eine Präsentation "Integration externer Normdatenquellen für Abgleich & Anreicherung lokaler Daten in #OpenRefine". https://t.co/Wz6L4JXs8P #reconciliation
Hier noch ein kurzfristiger Veranstaltungshinweis für #wikibase und #linkedData Interessierte
"NFDI-InfraTalk: Wikibase and the challenges and possibilities of knowledge graphs for RDM in NFDI4Culture"
Heute, 7.3.2022, 16 Uhr im Live Stream auf https://t.co/ub241O7CBC
#nfdi
Graphs of architecture documentation now in #Mermaid format (https://t.co/fgVNntMv8L) inside documentation (markdown) inside the Git(hub) repo, so all can edit them like other parts of the docs in https://t.co/yxlJ1riQGH - Tnx to #opensource#MkDocs plugin https://t.co/oMSuJ3YqIW
Working on more prioritization features for the task queue for better chance to process more relevant documents earlier. If you want to contribute on setting priorities for filename extensions or relevant keywords which often occure in filenames: https://t.co/6sXxbvNhYH #ddj
Most #opensource contributors of #Open Semantic Search not listed on Github user interface as "Contributors" because our repo is structured by Git submodules (additional git repos). Added a section "Contributors" to https://t.co/kTZtzhkdh8 (feel free to extend). Thanks to all!
Next Open Source release of Open Semantic Search Server with automatic setup of Celery Flower (web user interface) for monitoring of the document processing task queue (ETL) out of the box.
Working on #spaCy NER plugin to run Named-entity recognition (NER) by multiple different #MachineLearning models for same document language (currently "only" one #ml model per language configurable) to fill faceted search/interactive filters. #opensource#textmining#nlp#nlproc