What is a Confidence Score in RChilli?

In the context of RChilli's parsing and data extraction APIs, a Confidence Score represents the parser's level of certainty in the accuracy of the extracted information. It is a numerical value assigned to specific fields (like names, addresses, phone numbers, skills, job titles, etc.) in the parsed JSON output.

How the Confidence Score Works

• Scale: Typically ranges from 0 to 10.

• Higher Score (e.g., 9 or 10): Indicates strong certainty that the extracted data is accurate.

• Lower Score (e.g., 3 or 4): Suggests low confidence—manual review might be needed.

For example, when a resume includes:

{
"PhoneNumber": [
{
"Number": "+1845-305-8762",
"Type": "Phone",
"ConfidenceScore": 10
}
]
}

This means the parser is highly confident that the phone number was extracted correctly.

Where Confidence Scores Appear

• Contact Information: Phone, email, address

• Skills and Experience: Particularly when reqskillsdrill or reqexperiencedrill is enabled

• Job Titles and Education: Based on context and section formatting

Why Confidence Scores Matter

1. Quality Assurance: Allows systems to flag uncertain data for review.

2. Smart Filtering: Helps improve AI decision-making by focusing on high-confidence fields.

3. User Experience: Enables clients to provide feedback loops or override low-confidence data entries.

Example Use Cases

• In ATS platforms, confidence scores help auto-validate candidate data before saving it.

• In bulk resume imports, low-score entries can be sent for manual QA.

• In chatbot integrations, only high-confidence data is used for personalized interactions.

For Further Reading

You can refer to dynamic parser settings and sample API responses including confidence scores in the Resume Parser API documentation.

Feel free to contact us at support@rchilli.com if you have any additional queries.