AI Text Moderation, Toxicity & Sentiment Analyzer

About AI Text Moderation, Toxicity & Sentiment Analyzer

This API provides multilingual text analysis covering sentiment polarity, toxicity detection, emotion recognition, and aspect-based opinion extraction, along with an integrated content moderation layer. It offers two REST endpoints: /analyze.php for full analysis combining sentiment, toxicity, emotion, and aspect detection (with Full and Light detail levels), and /moderate.php, a lighter endpoint focused on real-time policy checks for hate speech, bullying, violence, sexual or graphic content, and spam. Requests accept single text strings or batches of up to 10 items, with configurable moderation settings including enforcement mode (soft or hard), sensitivity level, and per-category thresholds.

Toxicity detection covers categories such as hate speech, insults and bullying, threats or violence, self-harm, sexual content, graphic violence, and spam, each with individual confidence scores and optional evidence spans. Emotion recognition identifies primary emotions including joy, trust, anger, sadness, fear, disgust, and surprise. Aspect-based analysis extracts sentiment tied to specific entities or topics mentioned in the text, such as isolating opinions about "shipping" separately from overall sentiment. The API processes text in real time without storing or logging data, and works across multiple languages without additional configuration.

It is aimed at developers building customer feedback analysis, chat and community moderation, e-commerce review processing, media comment monitoring, brand sentiment tracking, and gaming or social app safety features. Three subscription plans are available: Basic at $0.00/month, Pro at $9.00/month, and Ultra at $39.00/month.

Key features

• Multilingual sentiment analysis with positive/neutral/negative/mixed polarity scores
• Toxicity detection across hate speech, harassment, violence, self-harm, sexual/graphic content, and spam
• Emotion recognition for joy, trust, anger, sadness, fear, disgust, and surprise
• Aspect-based analysis extracting opinions about specific entities or topics
• Full and Light analysis modes for depth vs. speed tradeoffs
• Configurable moderation with soft/hard enforcement and adjustable sensitivity thresholds

Frequently asked questions