Skip to content

Model blocks output structure

The JSON output of a model block varies according to the block type, the model inner workings and the block parameters.

The structure below corresponds to the maximum possible output, which can be obtained by turning on all output related functional parameters of the block.
When it is not an ML model placed in the workflow in advanced mode, the richness of the output also depends on the NL Core version of the model.

{
    "document": {
        "content": analyzed text (string),
        "language": language code (string),
        "version": technology version info (string),
        "categories": [],
        "entities": [],
        "extractions": [],
        "extraData": [],
        "knowledge": [],
        "mainLemmas": [],
        "mainPhrases": [],
        "mainSentences": [],
        "mainSyncons": [],
        "options": {},
        "paragraphs": [],
        "phrases": [],
        "relations": [],
        "sections": [],
        "segments: [],
        "sentences": [],
        "sentiment": [],
        "tags": [],
        "tokens": [],
        "topics": []
    }
}

All the keys are described in detail in the next page.
Use the links below to jump to the description of a specific key: