AI ALGORITHM FOR SKILLS EXTRACTION FROM RECOGNITION MESSAGES
A big corporation using recognition HR platforms for employee engagement wanted to use the recognition messages as a data source to get to know their employees better.
The client needed an algorithm to extract the soft and hard skills an employee is recognized for. The company generates approximately 1M of messages throughout the year.
We created a State-of-the-art NLP solution able to extract skills and phrases from unstructured data sources like recognition messages and create an employee portfolio based on them. This solution utilizes a top-notch Google-approved Natural Language Processing (NLP) model.
The ML model was trained to replicate the reasoning behind the dataset of manually prepared evaluations of employee appreciation messages. By understanding the reasoning behind these evaluations, the model aims to understand the context of each sentence and recognize when a part of that sentence tends to describe a soft or hard skill written in a message.
We created a production-ready solution able to detect soft skills and hard skills from 1M messages in less than 2 hours. The client has been using this solution as part of their HR platform since 2021.