Skip to content

dojodemos/Customer-Service-Conversational-Insights

 
 

Repository files navigation

Customer Service Conversational Insights

Knowledge Mining applied to Conversations

Every day a large amount of data is generated by Customer Services in the form of Natural Language conversations with customers. This data is stored as chat logs (with human agents or bots) or audio recordings from voice-based calls.

This is typically unstructured data that, if processed, can result in a great amount of insights for Contact Centers, add additional data to NBA models and provide a better view of the customer.

The Customer Service Conversational Insights is a solution accelerator built on top of Azure Cognitive Search, Bot Framework, Azure Speech Services and Text Analytics that allows customers to quickly deploy an integrated platform and immediately start extracting insights from customer conversations.

Demo site is build on hotel and travel booking conversations. Search for city name to get started (e.g. seattle, milan, etc...)

One click Azure Deployment

Audio data

To deploy the full solution with audio files transcription and the Web Application, select the following button:

Deploy to Azure

The Azure portal displays a pane that allows you to easily provide parameter values. The parameters are pre-filled with the default values from the template.

Once the deployment is completed, start processing your audio files by adding them in the "audio-input" container in the "storage account" in your "Resource Group". To navigate the Web Ui, check the "App Service" resource in your "Resource Group".

The template builds on top of the Ingestion Client for Speech service. Please check here for detailed parameters explainations: Getting started with the Ingestion Client

Architecture

alt text

Bot Telemetry with sample data

To deploy the full solution with Bot sample data and the Web Application, select the following button:

Deploy to Azure

The Azure portal displays a pane that allows you to easily provide parameter values. The parameters are pre-filled with the default values from the template.

Once the deployment is completed, start processing your Bot telemetry by exporting in the "conversational-km-raw" container in the "storage account" in your "Resource Group". To navigate the Web Ui, check the "App Service" resource in your "Resource Group".

OPTIONAL: You can also download a sample dataset here Unzip it and upload it in the "conversational-km-raw" container in the "storage account" in your "Resource Group".

Architecture

alt text

About

No description, website, or topics provided.

Resources

License

Code of conduct

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • CSS 76.7%
  • C# 8.6%
  • JavaScript 8.4%
  • Python 3.6%
  • HTML 2.5%
  • Dockerfile 0.1%
  • Shell 0.1%