Executable Request Classification
Executable Request Classification is the process of converting a natural-language user request into a structured decision that guides question understanding, knowledge scope, and answer depth.
Executable Request Classification is the process of converting a natural-language user request into a structured decision that guides question understanding, knowledge scope, and answer depth.
SonaMinds does not treat user input as something that should immediately enter retrieval or generation. A natural-language request must first be interpreted as a structured question. The system needs to understand what kind of question it is, what knowledge scope is relevant, what depth of answer is appropriate, and whether clarification is needed.
This is why SonaMinds uses the phrase Executable Request Classification. The classification is not merely a label such as “FAQ” or “analysis.” It is a structured decision about how the question should be handled before an answer is produced.
Why intent labels are not enough
Many systems use intent classification to assign a user request to a category. This is useful, but insufficient. The same words can require different behavior depending on context. A question asked on a course sales page may be a sales question. The same question asked in a lesson page may require guided tutoring. A request made in an expert workspace may require deeper analysis, while a similar request in a public setting may require a narrower answer or clarification.
Executable Request Classification therefore depends not only on the wording of the question, but also on the context in which the question is asked. The system should consider the relevant knowledge scope, the expected depth of response, and the conditions under which the answer is being requested. The result is a structured decision that can guide retrieval and answer generation.
What the classification controls
A classification result may guide the expected response type, the relevant knowledge scope, the depth of answer, and whether clarification is needed. These controls matter because they protect answer quality and keep the answer aligned with the intended context.
In SonaMinds, classification is best understood as the entrance to the answering system. It compresses the user’s natural language into a structured decision. This decision then guides retrieval, context selection, and answer behavior.
Conceptual boundary
Executable Request Classification is not the final answer. It does not generate the response. It decides how the response should be generated. This separation is important. If classification and answering are collapsed into one step, the system may answer before it has identified the proper knowledge scope, conceptual frame, or response depth.
The concept also distinguishes SonaMinds from a simple chatbot interface. The visible conversation may feel natural, but the answer still needs to be guided by structured interpretation.