Topic modeling determines the primary and secondary call drivers in contact center conversations. These call drivers are called topics.
Conversations
A conversation is an interaction between a contact center agent and an end user. Topic modeling analyzes conversations in the form of chat or voice call transcripts created using the Insights API.
For more information, see the Conversations reference documentation.
Topics
A topic describes the main subject discussed in a group of conversations, also called a primary call driver. Topic modeling creates topics by analyzing the key subjects from each of your conversations and creating clusters of similar subjects. Topic modeling then identifies the number of distinct clusters and attempts to generate a name for each one. Names represent a topic which in turn is represented by an Issue resource.
When topic modeling creates a set of topic names, you can review the names and the conversations it has labeled with that name. Topic modeling can also show you the snippet from the most representative conversation for a topic.
Secondary topics
Topic modeling can also identify secondary topics, which might be less interesting from an analysis perspective. Secondary topics are frequently related to regular process steps that happen in a conversation, such as authentication, confirmation, and collecting feedback.
Secondary topics can sometimes crowd out the more interesting primary topics, which makes it harder to spot the primary ones.
Topic models
The first step to identifying the topics in conversations is to create a topic model in Conversational Insights. A topic model contains a list of topics based on a group of conversations. From a topic model, you can generate a report identifying the topics within the model as well as the names and descriptions of each topic.
Topic models are represented by issueModels resources.
You can perform the following operations on topic models:
- Deploy
- Fine-tune
- Undeploy
- Delete
Fine-tune a topic model
After you create a topic model, you can review and modify the list of topics to fine-tune your model. There are three main techniques for fine-tuning a topic model to improve future topic assignments:
- Add a new topic.
- Edit an existing topic's name and description.
- Remove an existing topic.
All these actions affect adjusted topic distributions.
When you perform any of these fine-tuning actions, a new analysis follows the updated topic list, which means the existing analysis is unchanged. To apply a new change to an existing analysis, follow the instructions in Topic modeling instructions.
Topic inference
You can deploy a topic model to your project, which lets you infer topics in real-time during a new conversation. After you deploy your topic model, you can analyze each new conversation or analyze a set of conversations in bulk. The analysis determines which topic from your list applies to each conversation.