Google Cloud Natural Language

Score the overall sentiment of a text document, returning a score from −1.0 (negative) to +1.0 (positive) and a magnitude indicating emotional intensity, with optional sentence-level breakdown.

Extract named entities — people, organisations, locations, events, dates, numbers — from text, with salience scores and Wikipedia links where available.

Combine entity extraction with sentiment scoring to understand how the author feels about each specific named entity in the text — not just the document overall.

Parse the grammatical structure of text, returning tokens with part-of-speech tags, dependency parse arcs, lemmas, and morphological features for each word.

Classify a document into one or more categories from Google's hierarchical content taxonomy, with confidence scores — useful for routing, tagging, and content organisation.

Detect harmful or inappropriate content in text, returning confidence scores across categories such as toxic, insult, profanity, threat, and sexually explicit content.

Run sentiment analysis, entity extraction, and syntax analysis over a document in a single API call — the most cost-efficient option when you need all three outputs from the same text.

Process multiple text documents through sentiment analysis in a single agent run, returning scores and magnitudes for each input — ideal for end-of-period batch reporting.

Retrieve entity salience scores from an Analyze Entities call to identify which named entities are most central to the document's meaning — useful for keyword extraction and topic weighting.

Identify the BCP-47 language code of the input text using the Natural Language API's automatic language detection, returning the detected language and confidence level.

Extract the per-sentence sentiment breakdown from an Analyze Sentiment response, surfacing which individual sentences in a document are most positive or negative.

Classify text content into Google's taxonomy categories and apply a configurable routing rule to send the document or data to the correct downstream system or queue.

Filter the entities returned by Analyze Entities to return only PERSON-type entities, enabling targeted person-name extraction for CRM enrichment, compliance checking, or media monitoring.

Filter the entities returned by Analyze Entities to return only ORGANIZATION-type entities, isolating company and institution names from a document for prospecting, risk, and partnership tracking.