Profanity Detection - JigsawStack

Overview

The Profanity Detection API helps you identify and filter inappropriate language, slurs, and offensive content in text. Designed for content moderation across various platforms, our API goes beyond simple keyword matching to understand context and detect obfuscated profanity. Key Benefits:

High-accuracy detection of explicit and implicit profanity
Context-aware analysis that understands intent
Detection of obfuscated profanity and character substitutions
Multiple language support
Customizable sensitivity levels
Optional content masking

API Endpoint

POST /v1/validate/profanity

Quick Start

Javascript

import { JigsawStack } from "jigsawstack";

const jigsaw = JigsawStack({ apiKey: "your-api-key" });

const response = await jigsaw.validate.profanity({
  "text": "This is a sample text that might contain inappropriate language."
})

Response Example

{
  "success": true,
  "profanities_found": false,
  "profanities": [],
  "clean_text": "This is a sample text that might contain inappropriate language.",
  "_usage": {
        "input_tokens": 20,
        "output_tokens": 35,
        "inference_time_tokens": 235,
        "total_tokens": 290
  }
}

Use Cases

Content Moderation

Filter user-generated content on social platforms, forums, and comments sections.

Child Safety

Create safe environments for child-friendly applications and educational platforms.

Brand Protection

Ensure marketing content and customer interactions maintain brand-appropriate language.

Example Applications

Content Moderation Pipeline

Implement a complete content moderation workflow:

// Example: Complete content moderation pipeline
async function moderateContent(userContent) {
  // Step 1: Check for profanity
  const profanityResult = await jigsaw.validate.profanity({
    text: userContent,
    censor_replacement: "*"
  });
  
  // Step 2: If content has profanity, use the censored version
  let processedContent = profanityResult.profanities_found > 0 
    ? profanityResult.clean_text 
    : userContent;
  
  // Step 3: Add flags for content that may need manual review
  const needsReview = profanityResult.profanities_found > 3;
  
  // Step 4: Analyze sentiment on clean content
  const sentimentResult = await jigsaw.ai.sentiment({
    text: processedContent
  });
  
  return {
    original: userContent,
    processed: processedContent,
    contains_profanity: profanityResult.profanities_found > 0,
    profanity_count: profanityResult.profanities_found,
    sentiment: sentimentResult.sentiment,
    needs_review: needsReview
  };
}

Chat Application Filter

// Example: Real-time chat message filter
function ChatApp() {
  const [message, setMessage] = useState("");
  const [isSending, setIsSending] = useState(false);
  
  const handleSendMessage = async () => {
    if (!message.trim()) return;
    
    setIsSending(true);
    
    try {
      // Check message for profanity before sending
      const profanityCheck = await jigsaw.validate.profanity({
        text: message,
        censor_replacement: "*"
      });
      
      if (profanityCheck.profanities_found > 3) {
        // High profanity count - don't send and alert user
        alert("Your message contains inappropriate language and cannot be sent.");
        return;
      } else if (profanityCheck.profanities_found > 0) {
        // Contains some profanity - confirm with user to send censored version
        const willSendCensored = confirm(
          "Your message contains some inappropriate language. " +
          "Would you like to send a filtered version instead?"
        );
        
        if (willSendCensored) {
          await sendMessageToChat(profanityCheck.clean_text);
        } else {
          return;
        }
      } else {
        // No profanity - send as is
        await sendMessageToChat(message);
      }
      
      // Clear input after sending
      setMessage("");
    } catch (error) {
      console.error("Error sending message:", error);
      alert("Failed to send message. Please try again.");
    } finally {
      setIsSending(false);
    }
  };
  
  return (
    // Chat UI components...
  );
}

FAQ

How effective is the profanity detection for non-English content?

The profanity detection API is most effective for English content, but also provides good coverage for major languages including Spanish, French, German, Italian, and Portuguese. For other languages, basic profanity detection is available but may not catch language-specific slang or culturally specific offensive terms.

Can the API detect deliberately obfuscated profanity?

Yes, the API is designed to detect common obfuscation techniques such as:

Character substitution (e.g., “sh!t”, “f**k”)
Letter omission (e.g., “sh-t”, “fck”)
Deliberate misspellings (e.g., “phuck”)
Zero-width space insertion

However, very creative or unusual obfuscation methods might occasionally bypass detection.

How can I reduce false positives?

If you’re experiencing false positives (benign words incorrectly flagged as profanity), consider these approaches:

Post-process the results to whitelist certain terms specific to your domain
Implement a user feedback system to identify false positives
Add a human review step for edge cases

Find more information on Profanity Detection API here

​Overview

​API Endpoint

​Quick Start

​Response Example

​Use Cases

​Content Moderation

​Child Safety

​Brand Protection

​Example Applications

​Content Moderation Pipeline

​Chat Application Filter

​FAQ

​How effective is the profanity detection for non-English content?

​Can the API detect deliberately obfuscated profanity?

​How can I reduce false positives?

Overview

API Endpoint

Quick Start

Response Example

Use Cases

Content Moderation

Child Safety

Brand Protection

Example Applications

Content Moderation Pipeline

Chat Application Filter

FAQ

How effective is the profanity detection for non-English content?

Can the API detect deliberately obfuscated profanity?

How can I reduce false positives?