# Query Response
When you test a query in the query side bar, you normally get a response and information about intent and slot values.
If you click on the icon
at the lower left corner of the query side bar, you will get a full version of the response.
Here we are breaking them into more consumable groups and explain each field:
# Meta Data: (Informational Fields)
| Keys | Explanation |
|---|---|
| latencies | Dictionary of latencies at each step of the request. |
| latencies: total | The total latency of the request |
| latencies: traces | A breakdown of total latency into the latency for each micro service to help debug and optimize latency |
| latencies: webhook_total | Latency of the Business Logic request |
| latencies: webhook_latency | Dictionary of latency of database requests from external webhooks. Empty if not applicable. |
| latency | Same value as latencies: total |
| qid | UUID of query |
| today | Time stamp of the start of the query |
| version | Release version |
# Session Data: (Unchanging Fields For Entirety of Session)
| Keys | Explanation |
|---|---|
| institution_id | System generated identifier for the institution containing the AI version |
| session_id | System generated system managed identifier to track user sessions. Primarily used for logging and analytics. |
| user | UUID for request user |
# Query Data: (Parameters Used in Request)
| Keys | Explanation |
|---|---|
| ai_version | System generated identifier for the AI version that a query is hitting. |
| device | User provided device information from query end point |
| dialog | Dialog token passed back and forth to retain a session. If not passed back within 10 minuets, the session and dialog will be terminated. |
| external_user_id | A front end supplied ID that is used to track individual users from a usage and analytics perspective. This can be in any ASCII format up to 64 characters. Defaults to 1 unless supplied in query request. |
| lat | Latitudinal coordinate supplied in query request field of the same name |
| lon | Longitudinal coordinate supplied in query request field of the same name |
| query | User provided utterance from query end point |
| time_offset | UTC time offset supplied in query request field of the same name |
# NLU Results: (Data Fields Generated during NLU Pipeline)
| Keys | Explanation |
|---|---|
| response | Dictionary that contains data accumulated across the AI pipeline |
| response: query | User provided utterance from query end point |
| response: retained_slots | Intermediate slot data; list of slots carried over from the previous competency. Only used on cross competency transitions. |
| response: slots | Used to generate the slots field at root. See the Request Body: The slot dictionary for details. |
| response: subtype | The state in the competency |
| response: today | Time stamp of the start of the query |
| response: type | Same as the competency name |
| response: updated_slots | Intermediate slot data; slots that have been changed from the previous conversation state. |
| Keys | Explanation |
|---|---|
| classifier | The internal name of the classifier model |
| classifier_state | State of the competency the user is in currently |
| competency | Competency of the current conversation |
| competency_id | Internal ID representing the competency into which the query was classified. |
| consecutive_oos_number | consecutive_oos_number is used for calculating the number of consecutive out-of-scope queries. When classifier_state is in outofscope, consecutive_oos_number will add 1 to the value. A good usage for this field is to decide at what point you want live agent to take over after several out of scope responses. |
| emotions | Up to 5 emotion labels, each having its own probability score. It has two subfields: label, which is the emotion label, and score, which is the probability score of the emotion label. Refer to Emotion Labels for the complete list. |
| intent_raw | Intent classification used during AI pipeline |
| intent | Intent classification generated for AI pipeline |
| intent_probability | It’s a TRUE or FALSE variable based on the intent_probability score. It returns TRUE when the score is between |
| is_correction | System generated variable indicating a followup query (non root transition) in which a slot was retained. |
| is_repeat | System generated and deprecated flag. Default is false. |
| personality_id | Internal ID representing the response template used in generating the formattedResponse and speakableResponse. |
| sentiment | Sentiment of the query -1 negative, 0 neutral, 1 positive. |
| slots | Dictionary for extracted slots |
| slot_probabilities | Probability that an extracted slot is correct |
| slot_token_ranges | The start/end index of the slot token in the query |
# Response Data: (Fields Related to Response Generation aka NLP)
| Keys | Explanation |
|---|---|
| response_slots | response_slots is a top-level key added by AI developers that does not persist throughout the conversation. |
| visuals | Fields generated by the response template with rich response fragments. |
| visuals : formattedResponse | Response generated for response template and designated as a readable response. |
| visual_payload | Rich response used in personalities |
| visuals : speakableResponse | Response generated for response template and designated as a spoken response. |
# Emotion Labels
Given below is the complete list of emotion labels available in emotions['label']
- admiration
- amusement
- anger
- annoyance
- approval
- caring
- confusion
- curiosity
- desire
- disappointment
- disapproval
- disgust
- embarrassment
- excitement
- fear
- gratitude
- grief
- joy
- love
- nervousness
- optimism
- pride
- realization
- relief
- remorse
- sadness
- surprise
- neutral
Last Updated: 04/17/2024