NLP Pipelines

NLP systems are often built around a pipeline where text flows in and results are generated. The pipeline illustrated below is constructed using our NLP Building Blocks and it shows a pipeline to perform named entity extraction based on the text’s language and contents.

The text is first passed to Renku to determine the language of the text. Next, based on the language the text passes to the appropriate instance of Sonnet for tokenization. The type of document is determined by Lacuna, and based on the document type, the tokens are passed to an instance of Idyl E3 for named entity extraction. This pipeline uses all self-contained applications and can fully operate behind firewalls.

Use docker-compose to create the NLP building blocks.

More Information?

Use the form to the right to get in touch. We’ll be happy to discuss our applications with you in depth.