Obsei Workflow Execution via GitHub Action
This repo Showcase Obsei workflow execution via GitHub actions. We have created this for DataTalks.Club's demo.
This repo represent following workflow -
- GoogleNews Observer search and fetch google news article with given query
- Then TextSplitter split article in small chunks so Classification Analyzer can process it easily
- Then Classification Analyzer classify chunks in given labels using HuggingFace zero shot models
- Then Inference Aggregator aggregate classification output for each chunks and compute final inference based on given aggregation function
- And finally Slack Informer send result to given slack channel using API
Screenshot of final result -
- Full article text
- Final article classification result
- Fork this repo
- Update workflow.yml based on your need (Note: Never add you credentials in yml file, instead use environment variable and fetch them via github secret)
- [Optional] Update workflow.py if you would like to add some custom code
- Add required credentials in the secret to prevent their exposure to public
- Update github action.yml along with required environment variables (For example scheduling at regular time or based on some event refer link for more detail)
- Spread Obsei love with the world :)
This link might help you to regarding how collect credetials of observers (Facebook, Twitter etc) and informers (Slack, Zendesk etc).