Entity Recognition
What's being used in the demo ?
A deep learning architecture with Bi LSTM char level encodings 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 faster 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 Sentiment Analysis, can be used to robustly analyze
thousands of comments, posts, messages by customers in minutes.
• With finetuned NER analyzing and extracting useful and important data points from large texts
is very simple.
• Can be directly integrated into existing customer support system to increase the efficiency of customer support by identifying
the entities in the customer's queries.
