Obsei

Obsei is intended to be an automation tool for text analysis need. Obsei consist of -

  • Observer, observes platform like Twitter, Facebook, App Stores, Google reviews, Amazon reviews, News, Website etc and feed that information to,
  • Analyzer, which perform text analysis like classification, sentiment, translation, PII etc and feed that information to,
  • Informer, which send it to ticketing system, data store, dataframe etc for further action and analysis.

Current flow


Future concept (Coming Soon! ?)

Use cases

Obsei use cases are following, but not limited to -

  • Automatic customer issue creation based on sentiment analysis (reduction of MTTD)
  • Proper tagging of ticket based for example login issue, signup issue, delivery issue etc (reduction of MTTR)
  • Checking effectiveness of social media marketing campaign
  • Extraction of deeper insight from feedbacks on various platforms
  • Research purpose

Demo

We have a minimal streamlit based UI that you can use to test Obsei.

Screenshot

To test remotely, just open: Obsei Demo Link
(Note: Due to rate limit sometime Streamlit demo might not work, hence please use docker image locally.)

To test locally, just run

docker run -d --name obesi-ui -p 8501:8501 obsei/obsei-ui-demo

# You can find the UI at http://localhost:8501

Support and Release Matrix

Linux Mac Windows1 Remark
Tests Low Coverage as difficult to test 3rd party libs
PIP Fully Supported
Conda2 Partially Supported

On Windows you have to install pytorch manually. Refer Pytorch official instruction.

Conda channel missing few dependencies, hence install missing dependencies manually -
Missing Conda dependencies

pip install presidio-analyzer
pip install presidio-anonymizer
pip install zenpy
pip install searchtweets-v2
pip install google-play-scraper
pip install tweet-preprocessor
pip install gnews
pip install trafilatura
pip install python-facebook-api