To increase accuracy with minimal development effort, you should utilize Active Learning. Active Learning analyzes user queries that resulted in low-confidence matches or close proximity matches and suggests them as "alternative phrases" for existing question-and-answer pairs.

1. Open the Review suggestions pane: This is the specific interface in Language Studio where Active Learning suggestions are presented.
2. Review and accept the alternative phrases: You must explicitly accept the suggested phrasings to add them to your knowledge base, improving its ability to recognize those specific common user queries.
3. Retrain and republish the model: Changes to the knowledge base (like adding new phrasing) are not live until the model is saved/trained and republished (deployed) to the endpoint.

Microsoft Learn - Azure AI services: "Accept active learning suggestions" - This article details the workflow of selecting the Review suggestions icon (Step 1), filtering and accepting suggestions (Step 2), and then Saving/Deploying (Step 3).

Source: https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/how-to/active-learning

Microsoft Learn - Question Answering: "Review suggestions" - "Active learning suggestions are displayed in the Review suggestions pane... After you accept or reject suggestions... select Deploy knowledge base."

Source: https://learn.microsoft.com/en-us/azure/ai-services/language-service/question-answering/concepts/active-learning