10 March 2023
Rating Report Automation based on Natural Language Generation

Client Profile

Client is a Mumbai based Credit Rating Agency which provides the entire spectrum of credit rating. This case study showcases how our platform uses the Natural Language Generation (NLG) to generate the Rating Reports and Rationale Reports for the client.


To use the AI technique to improve report writing.
To assist analysts by automating their entire report generation processes.

Problem Statements

There were numerous industries & thousands of borrowers associated to the industry which was served by our client.
Creating industry specific corpus of words & sentence as we can’t have same types of sentences/words for all industries.
Creating dynamic & industry specific templates for rationale report, rating report & press release.
Creating industry specific NLG models and validating the output against each industry.


The Solution was a Natural Language Generation technique to generate the Press Release, Rating Report & Rationale Report automatically for one of the largest Credit Rating Agencies in India.
Uploaded the corpus of words & sentences for various industries.
Created various models and grouping them based on industry’s homogenous characteristics.
Templatized the reports such as rationale report, rating report & press release.
Used models from probabilistic classifiers, deep learning and natural processing to transform the corpus of sentences into a meaningful paragraph.
Trained the models & compared the outputs with the old reports & press releases.
Generated the reports automatically and send it for the analyst’s review.


In the first phase, the client was able to automate their entire report generation processes by 60% with accuracy of around 75%.
With multiple rounds of iteration and periodic reviews of the models done, based on analysts’ reviews and hyper parameter tuning and corpus updates, we were able to get the model accuracy of 85%.