Sentiment Classification
What's being used in the demo ?
A custom deep learning architecture along with state of art BERT
embeddings for transfer learning.
We also employ a simple neural network based model for some tasks
which can train on millions of data points in under a minute and achieve comparabale with SOTA results.
Unlike most deep learning architectures, our models can run inference at great speed, even without GPU.
Applications
• Can be directly integrated with bots to provide intelligent responses to customer queries.
• Combined with NER, can be used to robustly analyze
thousands of comments, posts, messages by customers in minutes.
• With finetuned Text Classification analyzing and extracting useful and important sentences from large texts
is very simple.
• Can be directly integrated into existing customer support system to increase the efficiency of customer support by provinding
pre-defined resposnes for common customer queries.
