In Azure Cognitive Service for Language (Question Answering), when users ask natural language questions, the service tries to match them to knowledge base content and synonyms.
In the scenario, the model correctly answers "What is the price of < product > ?" but fails on "How much does < product > cost?" .
This indicates that the service does not automatically recognize "price" and "cost" as equivalent terms.
Creating an entity for cost will not fix this issue. Entities in custom text classification or conversational language understanding are used for extracting structured data (like dates, names, or locations) from text, not for synonym matching in question answering.
The correct solution is to add alternate phrasing (synonyms/alternate questions) in the QnA project for each knowledge base entry. For example, under the QnA pair that contains "What is the price of < product > ?" , you would add "How much does < product > cost?" as an alternate question. This way, both variations trigger the same answer.
Therefore, the proposed solution (create an entity for cost) does not meet the goal.
The answer: B. No
Microsoft References
Add alternative questions to improve matching - Azure AI Language Question Answering Improve accuracy with synonyms and alternate phrasing Overview of Question Answering in Azure AI Language