Training Flow for Unhandled Phrases on Virtual Assistants - OvationCXM - Product Updates

Training Flow for Unhandled Phrases on Virtual Assistants

Hello, OvationCXM world!

We are thrilled to announce the release of the Topic Modeling Flow for training unhandled responses in the OvationCXM Virtual Assistant. Unhandled responses are phrases that your customers type when interacting with a Virtual Assistant that don't match to a response you have defined.

The following feature is now available under the clustering section on the VA navigation panel. This innovative feature empowers users to enhance the VA's understanding and response accuracy by leveraging advanced clustering and modeling techniques to efficiently organize and train unhandled conversations.

Overview:

The Topic Clustering Flow is a powerful feature that empowers users to efficiently train unhandled and missed user responses. With this feature, our state-of-the-art AI embedding model will create clusters of unhandled phrases for users to map them to intents, and train the system for improved response accuracy.

The Flow:

Screenshare - 2023-05-19 1_58_33 PM.mp4 - Google Drive

Loading…

Key Features:

• Cluster Creation: The Topic Modeling Flow automatically generates and names the clusters of unhandled phrases. By clicking on "Populate Clusters," users can quickly view and organize unhandled responses based on common topics and patterns. Users can also manually create new clusters to categorize relevant unhandled phrases into it.

• Cluster Details and Metrics: Users can access a dedicated view that provides comprehensive information about each cluster. This includes overall impact metrics, top keywords, the status of each cluster, and the intent it is mapped to. This overview helps users understand the relevance and importance of each cluster.

• Phrase Acceptance and Rejection: Users can review the phrases within each cluster and accept or reject them accordingly. Accepting a phrase indicates that it should be mapped to the associated intent while rejecting a phrase excludes it from the training process and maps it to an ignored phrases pool.

• Cluster Management: Within the cluster view, users have the flexibility to perform various actions. They can rename clusters, map them to specific intents, and interact with the phrases within each cluster.

• Drag and Drop Functionality: To further customize the clusters, users can easily drag and drop phrases from one cluster to another. This feature allows for intuitive organization and refinement of the unhandled responses.

• Training Process: Once all the clusters are prepared, users can initiate the training process. This step ensures that the phrases within each cluster are properly mapped to the corresponding intents. After training, the system is reset and ready for the next round of virtual assistant training.

Benefits:

• Enhanced Training Accuracy: By mapping phrases to specific intents and training the system, users can significantly improve the Virtual Assistant's response accuracy. This leads to a more seamless and satisfactory user experience.
• Efficient Cluster Management: The Topic Modeling Flow simplifies the organization and management of unhandled responses by automatically creating clusters based on similar topics. Users can easily navigate and interact with clusters, saving time and effort.
• Intuitive User Interface: The drag-and-drop functionality and user-friendly cluster view make it straightforward for users to customize and refine the clusters according to their requirements.
• Streamlined Training Process: The automated training process ensures that phrases are accurately mapped to intents, eliminating manual intervention and reducing the risk of errors.

We are excited to introduce the Topic Modeling Flow in the OvationCXM Virtual Assistant and we would appreciate your feedback as we continue to enhance our VA platform.

For more information or assistance, please contact our product team at product@ovationcxm.com.

Note: This release note provides a high-level overview of the Topic Modeling Flow features. For detailed instructions on how to utilize this feature, we will be adding relevant knowledge articles and documentation.

Vaibhav Gupta