BatchPredict

Predicts the intents associated with a batch of utterances.

Path Parameters
namespace string — REQUIRED Namespace where the workspace is located.
playbook_id string — REQUIRED Unique identifier of the workspace whose model is to be used.

Request body — REQUIRED
namespace string Namespace where the workspace is located.
playbook_id string Unique identifier of the workspace whose model is to be used.
input_utterances string[] Utterances to classify.
model_id string Optional. Specifies which model to use through its unique NLU engine identifier. If none specified, the default is used.
revision_id string Optional. Specifies which revision of the model to use.
intent_tags TagPredicate require_ids string[] Only include objects with ALL of the given tag ids. include_ids string[] Only include objects with ANY of the given tag ids. exclude_ids string[] Exclude objects with ANY of the given tag ids.

Response Body

predictions PredictResponse[] Predictions for each utterances given in the request. The order is guaranteed to be the same as in the request. matches IntentMatch[] Raw list of matches ordered by descending score (no hierarchical considerations) id string The intent id that was matched. name string The name of the intent when the model was trained. hierarchy_ids string[] Intent IDs of the lineage of the matched intent. The first ID is the root parent intent, the last is the matched intent. hierarchy_names string[] Intent names of the lineage of the matched intent. The first name is the root parent intent, the last is the matched intent. score float The probability of this being the right matched, as determined by the underlying model. hier_matches HierIntentMatch[] List of matches rescored with hierarchical considerations. id string The intent id that was matched. name string The name of the intent when the model was trained. score float The recursive sum of the scores of all the sub-intents. own_score float The probability of this intent, if its children were disregarded. children_entropy float The entropy of the intent's direct children. children HierIntentMatch (circular)[] The match probabilities for this intent's direct children entity_matches EntityMatch[] Entity matches, if available entity EntityReference Sync from zia.ai.playbook.EntityReference entity_id string Unique identifier of the entity for database storage. key string Key by which we can refer to this entity in the utterance. Ex (rasa): I'd like to visit [New York City](city) where city is the key text string Text used to reference the entity in the utterance Ex (rasa): I'd like to visit [New York City](city) where New York City is the text. value string If the reference text isn't the main entity value, this value points to the right key value to use. For example, for a city entity, if a synonym was used, this value would contain the key value it refers to in the entity. (rasa long): I went to NYC{"entity": "city", "value": "New York City"} where 'New York City' is the value (rasa short): I went to NYC(city:New York City) value_id string Unique identifier of the entity value for database storage. role string If entity has repeated usage in the utterance, assigns role for each usage Ex (rasa): I want to fly from Berlin{"entity": "city", "role": "departure"} to San Francisco{"entity": "city", "role": "destination"}. score float The confidence score for this match span Span start uint32 The start of this entity, as the utf8 byte index end uint32 The end of this entity, as the utf8 byte index extractor string (Rasa specific) Name of the pipeline component that found this entity parts InputPart[] If there are entity matches, matches in part format for convenience text Text text string entity EntityMatch entity EntityReference Sync from zia.ai.playbook.EntityReference entity_id string Unique identifier of the entity for database storage. key string Key by which we can refer to this entity in the utterance. Ex (rasa): I'd like to visit [New York City](city) where city is the key text string Text used to reference the entity in the utterance Ex (rasa): I'd like to visit [New York City](city) where New York City is the text. value string If the reference text isn't the main entity value, this value points to the right key value to use. For example, for a city entity, if a synonym was used, this value would contain the key value it refers to in the entity. (rasa long): I went to NYC{"entity": "city", "value": "New York City"} where 'New York City' is the value (rasa short): I went to NYC(city:New York City) value_id string Unique identifier of the entity value for database storage. role string If entity has repeated usage in the utterance, assigns role for each usage Ex (rasa): I want to fly from Berlin{"entity": "city", "role": "departure"} to San Francisco{"entity": "city", "role": "destination"}. score float The confidence score for this match span Span start uint32 The start of this entity, as the utf8 byte index end uint32 The end of this entity, as the utf8 byte index extractor string (Rasa specific) Name of the pipeline component that found this entity model_id string The model that was used to make this prediction. revision_id string The revision of the model that was used to make this prediction.