false positive proprietary license detected

[AyanSinhaMahapatra]

For the following text we detect a proprietary license which is not true: Hint.css is free for personal and commercial use under the MIT License.

[levi42x]

Test Result

• Scanning the text: Hint.css is free for personal and commercial use under the MIT License.
• Produced license expression: proprietary-license AND mit
• Expected result: mit

Root Cause

File : [proprietary-license_272.RULE]

• Matches the phrase "Commercial Use"
• Marked as is_license_clue: yes

File : [mit_88.RULE]

• Matches "under the MIT license"
• Marked as is_license_notice: yes

The Detection Flow for Our Text

Input: "Hint.css is free for personal and commercial use under the MIT License"
         ↓
    Tokenization
         ↓
    Aho-Corasick Matching
         ↓
    ┌──────────────────────────────────────────┐
    │  Match 1: "commercial use"               │
    │  Rule: proprietary-license_272.RULE      │
    │  Expression: proprietary-license         │
    │  Relevance: 100                          │
    │  is_license_clue: yes                    │
    ├──────────────────────────────────────────┤
    │  Match 2: "under the MIT License"        │
    │  Rule: mit_88.RULE                       │
    │  Expression: mit                         │
    │  Relevance: 100                          │
    │  is_license_notice: yes                  │
    └──────────────────────────────────────────┘
         ↓
    Group Matches (same line)
         ↓
    group_of_matches = [Match1, Match2]
         ↓
    Analyze Detection
         ↓
    is_false_positive() check
         ↓
    Both have relevance=100 → NOT flagged as false positive
         ↓
    Create LicenseDetection
         ↓
    Combine expressions: proprietary-license AND mit
         ↓
    ❌ WRONG RESULT

Proposed Fix :

1. Decrease relevance of [proprietary-license_272.RULE] so it does not override clear license declarations.
2. Modify is_false_positive() check.
3. Rule modification
4. any other suggestion : please...

[Adityatorgal17]

Is this issue still open?
@AyanSinhaMahapatra @pombredanne

[levi42x]

hi @Adityatorgal17,
i did a bit of a deep dive into the root cause already—check out the Proposed Fix section.

If you have any other ideas or a more optimal way to fix the false positive, let me know! we can discuss it here before you start on the code.