Use Case Identification from Chat Logs

Use case recommendation is a key feature to identify, classify and extract relevant intents from chat logs and user chat data. This technique is vitally useful in the following scenarios:

  • Extracting intents from ticketing data
  • Extracting intents from failed utterances

BigFix AEX should be trained on a periodic basis on the intents extracted by this method to continuously reduce failed utterance and increase intent recognition over time. The model works by classifying utterances in separate clusters which can be further trained. Use case recommendation can be found in the use case design console.

The following steps show how use case recommendation can be used to classify intents from data dump:

  1. Navigate to use case design console of the BigFix AEX tenant from which data is to be extracted or new intents are to be setup.
  1. Select a particular skill then Go to the Intent tab where you will find a button called use case recommendation.
Figure 1. Figure 167 – Use Case Recommendation in UCD
  1. Click on the “USECASE RECOMMENDATION” button, which will open a modal/popup in the same window.
Figure 2. Figure 168 – Use Case Recommendation Modal
  1. There are three stages available in the modal:
  1. Source: Selection of source of the data
  1. Chat log: Providing data to the recommender
  2. Recommendation: Fetching Recommendations
  3. Choose Source for dump to get recommendations, you can select as Upload CSV or Suggest use-cases, which provides recommendations from the current BigFix AEX instance.