Extract Named Entity Recognition Outputs from Tasks
You might already know that Named Entity Recognition (NER) tasks (an NLP technique to identify and classify key information in a body of text) are available in Composable ML. You can freely swap out and add the SpaCy NER task (currently the only available NER task as of 7.3) to create your own custom blueprint. This can be easily done by modifying the processing tasks associated with your text variables. Outputs from the NER task are directly used in the next processing task.
However, did you know that you can use a custom task with Composable ML to extract the outputs from the NER task? Simply start by adding a custom task after your NER task.
From here, you can provide your own custom code in R or Python to extract and save the NER outputs to a CSV file. You can make the CSV file available for download from ArtifactDownload.
If you have questions about this tip, just let me know by replying here to my post or you can send me a DM @jchang1161. I’d love to hear how you’re using Composable ML or NER!