Skip to content

Issue with malyalam and kannad normalization, extra space is inserted before punctuation #2

Description

@raccoon641

I have tried normalizing malyalam and kannada but in the output was not able to normalize phone number
Input:
{"text":"എനിക്ക് 29 വയസുണ്ട്, എന്റെ ഫോൺ നമ്പർ 9123456789 ആണ്."}
Output:
{
"normalized_text": "എനിക്ക് ഇരുപത്തൊമ്പത് വയസുണ്ട് , എന്റെ ഫോൺ നമ്പർ 9123456789 ആണ് .",
"detected_lang": "ml",
"lang_name": "Malayalam",
"processing_time": 0.05313992500305176
}

Also I am putting together all observed issues accross langauges

Observed Behavior

  1. Phone number normalization (critical)
    Phone numbers are handled differently across languages:
    Hindi (hi): Digit-wise normalization but extra leading digit introduced
    Tamil (ta): Digit-wise normalization, but non-Tamil zero lexeme (பூஜ்யம்) used
    Telugu (te): Phone number treated as a cardinal quantity, expanded into crores/lakhs
    Kannada (kn): Phone number not normalized at all
    Malayalam (ml): Phone number not normalized at all

  2. Punctuation normalization (global bug)
    Across all tested languages, an extra space is inserted before punctuation:
    Examples:
    है ।
    ஆகும் .
    ಇದೆ .
    ആണ് .

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions