MeghanaKap/indian_names
indian_names Unique customer names (lowercased) extracted from call/loan record databases (xPertVoice Aug/Sep, both "PQ" and "PQ2" variants), each paired with the language(s) detected from s3_key. customer_names_with_language.csv — two columns: name, language. One row per unique name. If a name was seen under multiple languages, all of them are listed in one comma-separated field (e.g. "english, tamil"). Language is detected by scanning s3_key for a known Indian-language… See the full description on the dataset page: https://huggingface.co/datasets/MeghanaKap/indian_names.
indian_names
Unique customer names (lowercased) extracted from call/loan record databases (xPertVoice Aug/Sep, both "PQ" and "PQ2" variants), each paired with the language(s) detected from s3_key.
customer_names_with_language.csv— two columns:name,language. One row per unique name. If a name was seen under multiple languages, all of them are listed in one comma-separated field (e.g."english, tamil").- Language is detected by scanning
s3_keyfor a known Indian-language keyword (telugu, tamil, kannada, malayalam, hindi, english, ...) rather than assuming a fixed path position. unknownis used only when a name was NEVER seen with any real language tag (i.e. every occurrence had a missing/unmatcheds3_key). If a name has bothunknownand a real language somewhere in the data,unknownis dropped and only the real language(s) are kept.- Names are also whitespace-normalized (repeated spaces collapsed, trimmed) before deduplication, so e.g.
"a badrinath"and"a badrinath"merge into one row. - 3,032,119 unique names total.
Per-language presence (each name counted once per language it appears under):