AI Certification Parser Returns Too Many Unmentioned Certifications

Problem Description

The AI certification parser is returning excessive certifications that are not mentioned in the job listing, causing inaccurate job metadata.

Example Case

Input Job Listing Text:

• Salesforce Platform Developer I (Required)
• Salesforce Platform Developer II (Preferred)
• Additional certifications such as JavaScript Developer I or Salesforce Architect credentials are a plus

Current AI Output (Incorrect):

Platform Developer, JavaScript Developer, Application Architect, B2B Solution Architect, B2C Commerce Architect, B2C Solution Architect, Heroku Architect, MuleSoft Platform Architect, Platform Integration Architect, Platform Data Architect, Platform Development Lifecycle and Deployment Architect, Platform Identity and Access Management Architect, Platform Sharing and Visibility Architect, Platform Strategy Designer, System Architect, Tableau Architect

Expected Output:

Platform Developer I, Platform Developer II, JavaScript Developer I

Root Cause

The AI is interpreting 'Salesforce Architect credentials' as a broad category and returning ALL architect certifications instead of treating it as a general mention.

Proposed Solution

Update the certification parsing prompt to:

1. Only extract explicitly mentioned certifications
2. When generic terms like 'Salesforce Architect' are mentioned without specifics, do not expand to all possible architect certifications
3. Maintain distinction between required, preferred, and nice-to-have certifications
4. Return only the certifications directly referenced in the text

Impact

This issue affects job listing accuracy and may mislead job seekers about actual certification requirements.