Nlp Functions¶

Quick Reference¶

Function Details¶

INFERENCE¶

INFERENCE(endpoint_id STRING, input STRING, timeout STRING) -> STRING

Performs inference using a configured Elasticsearch inference endpoint.

Parameters:

Returns: STRING - The inference result as text

Examples:

INFERENCE('my-openai-endpoint', 'Explain Elasticsearch')
INFERENCE('elser-endpoint', 'search query', '60s')

INFERENCE_CHAT¶

INFERENCE_CHAT(endpoint_id STRING) -> STRING

Performs chat completion using the Unified Completion API (OpenAI-compatible).

Parameters:

Returns: STRING - The assistant's response text

Examples:

INFERENCE_CHAT('my-openai-endpoint', '[{\
:\
,\
:\

INFERENCE_EMBED¶

INFERENCE_EMBED(endpoint_id STRING, input STRING, timeout STRING) -> ARRAY

Generates embeddings for the input text using a configured embedding endpoint.

Parameters:

Returns: ARRAY - The embedding vector as an array of numbers

Examples:

INFERENCE_EMBED('my-embedding-endpoint', 'text to embed')
INFERENCE_EMBED('elser-endpoint', 'search query')

INFERENCE_RERANK¶

INFERENCE_RERANK(endpoint_id STRING, query STRING, documents ARRAY, top_n NUMBER, timeout STRING) -> ARRAY

Reranks a list of documents based on relevance to a query using a reranking endpoint.

Parameters:

Returns: ARRAY - Reranked documents with scores

Examples:

INFERENCE_RERANK('my-reranker', 'elasticsearch query', doc_array)
INFERENCE_RERANK('cohere-rerank', 'best database', docs, 5)

LLM_CHAT¶

LLM_CHAT(messages ARRAY OF DOCUMENT, model STRING, api_key STRING) -> STRING

Sends a chat conversation to OpenAI and returns the assistant's response.

Parameters:

Returns: STRING - The assistant's response

LLM_CLASSIFY¶

LLM_CLASSIFY(text STRING, categories ARRAY OF STRING, api_key STRING) -> STRING

Classifies the given text into one of the provided categories using an LLM.

Parameters:

Returns: STRING - The selected category

Examples:

LLM_CLASSIFY('Server is down!', ['critical', 'warning', 'info'])
LLM_CLASSIFY(email_body, ['spam', 'not_spam'])

LLM_COMPLETE¶

LLM_COMPLETE(prompt STRING, model STRING, api_key STRING) -> STRING

Generates a text completion using OpenAI's API. Returns the generated text.

Parameters:

Returns: STRING - The generated completion text

Examples:

LLM_COMPLETE('Explain Elasticsearch in one sentence')
LLM_COMPLETE('Translate to French: Hello', 'gpt-4o')

LLM_EMBED¶

LLM_EMBED(text STRING, model STRING, api_key STRING) -> ARRAY OF FLOAT

Generates an embedding vector for the given text using OpenAI's embedding API.

Parameters:

Returns: ARRAY OF FLOAT - The embedding vector

Examples:

LLM_EMBED('Elasticsearch is a search engine')
LLM_EMBED('Hello world', 'text-embedding-3-large')

LLM_EXTRACT¶

LLM_EXTRACT(text STRING, fields ARRAY OF STRING, api_key STRING) -> DOCUMENT

Extracts structured information from text based on a schema.

Parameters:

Returns: DOCUMENT - Document with extracted fields

Examples:

LLM_EXTRACT('John Doe, age 30, from NYC', ['name', 'age', 'city'])
LLM_EXTRACT(error_log, ['error_type', 'timestamp', 'severity'])

LLM_SUMMARIZE¶

LLM_SUMMARIZE(text STRING, max_words NUMBER, api_key STRING) -> STRING

Summarizes the given text using an LLM. Returns a concise summary.

Parameters:

Returns: STRING - The summarized text

Examples:

LLM_SUMMARIZE(long_document)
LLM_SUMMARIZE(log_output, 50)