Automatic Call Routing Leveraging Deep Learning
A case study using voice to text and NLP
The customer was looking to seamlessly and effectively manage public safety with next generation control centers and dispatchers. They required:
- Lower call center routing latency and lower Abandonment Rate
- Improve customer experience through customer/data-360 integration
- Automatically route call to correct agent
- Run deep learning models on transactional data in real time
- Continuous training models are built and deployed with no downtime
- Automatically trigger transactional workflows to route call to correct agent based on prediction criteria and scoring
- Leverage deep learning without GPU
With GigaSpaces’ InsightEdge, integrated with Intel’s BigDL and Xeon processors, customers can run real-time Deep Learning models on the incoming calls to automatically route them to best agent according to their need.
- The user speaks using web interface
- Browser converts speech to text and sends to controller
- Controller writes data to a Kafka topic
- Spark job listens on Kafka topic and using BigDL model, creates prediction
- BiGDL Prediction to InsightEdge
- InsightEdge event processor listens for Prediction data and routes call session
- Increase of ~30% performance over previous processors.
- Simplified ETL, reducing component and cluster sprawl for optimal performance and TCO.
- Classify calls, retrieve customer data and route to relevant agent in ~50 ms