Replace Stanford NLP with a less memory-intensive alternative

Onto-DESIDE-VSE / patomat2

PatOMat2 - A Tool for Pattern-Based Ontology Transformation using SPARQL

MIT License

0 stars 0 forks source link

Replace Stanford NLP with a less memory-intensive alternative #48

Closed ledsoft closed 1 day ago

ledsoft commented 1 month ago

StanfordNLP which we use for tokenization and other NLP tasks has high memory requirements. We need to optimize it or replace it with a less memory-intensive alternative.

ledsoft commented 1 day ago

Disabled some of the annotators. Memory footprint has been significantly reduced without affecting the functionality.