A neural network algorithm that extracts topics from texts and highlights the relevant sentences while providing rank of the topics; also featured tag-based search.
This project was to build a Natural Language Processing model to classify the reviews for cars on multiple Chinese social networks. By deploying this model on the company’s big data platform, the platform enables fully automated data collection, data processing, data analyzing, and reports generating to monitor public comments, improve production procedures, and give risk warning in an early stage.
Using a combination of algorithms like FastText, TD- IDF, and BERT, the model also involves self-designed space vector mapping to filter complaints from endorsements before generating ranked tags to categorize review subjects in a granular fashion. The algorithm could also self-optimize ranked tags as new data was collected and processed.