COMI-LINGUA (COde-MIxing and LINGuistic Insights on Natural Hinglish Usage and Annotation) is a large-scale, expert-annotated dataset for Hindi-English code-mixed text. It supports multiple foundational NLP tasks and is designed to benchmark code-mixed and multilingual models.

📝 Dataset Summary

• Languages: Hindi (hi), English (en)
• Tags: code-mixing, Hinglish, expert-annotated
• Size: 181,463 Instances annotated by three annotators.
• License: CC BY 4.0

📚 Annotation Tasks

1. Language Identification (LID)

Token-level classification of each word as hi, en, or other.

Based on Microsoft LID tool, reviewed and corrected by annotators.

Sentence: प्रधानमंत्री  नरेन्द्र  मोदी  डिजिटल  इंडिया  मिशन  को  आगे  बढ़ाने  के  लिए  पिछले  सप्ताह  Google  के  CEO  सुंदर  पिचाई  से  मुलाकात  की  थी ।
LID tags:    hi   hi   hi    en    en    en   hi  hi   hi  hi  hi   hi    hi    en    hi  en   hi   hi   hi   hi   hi  hi ot

---

2. Matrix Language Identification (MLI)

Identifies the dominant language in each sentence.

Sentence: "India’s automation and design expert pool is vast, और ज़्यादातर Global companies के इंजीिनयिंरग center भी भारत में हैं।"
Matrix Language: en
sentence:"किसानों को अपनी फसल बेचने में दिक्कत न हो इसके लिये Electronic National Agriculture Market यानि ई-नाम योजना तेजी से काम हो रहा है।"
Matrix Language: hi

---

3. Part-of-Speech Tagging (POS)

Token-wise part of speech tags using the CodeSwitch library, then refined by annotators.

Sentence:  भारत   द्वारा  बनाया  गया Unified Payments Interface  यानि   UPI    भारत   की    एक   बहुत  बड़ी   success  story  है    ।
POS tags: PROPN  ADP VERB VERB  PROPN   PROPN    PROPN    CONJ  PROPN  PROPN ADP  DET  ADJ  ADJ    NOUN    NOUN  VERB  X

---

4. Named Entity Recognition (NER)

Token-wise entitiy recognition such as persons, location, organizations, gpe, and date using the CodeSwitch library then corrected by annotators.

Sentence: "मालूम हो कि पेरिस स्थित Financial Action Task Force, FATF ने जून 2018 में पाकिस्तान को ग्रे लिस्ट में रखा था।"
NER Tags: "पेरिस" → GPE, "Financial Action Task Force, FATF" → ORGANISATION, "2018" → DATE, "पाकिस्तान" → GPE

---

5. Machine Translation (MT)

Parallel translations for each sentence into:

• English
• Romanized Hindi
• Devanagari Hindi

Generated using Llama 3.3 LLM, then refined manually.

Sentence: भारत में भी green growth, climate resilient infrastructure और ग्रीन transition पर विशेष रूप से बल दिया जा रहा है।
English: In India too, special emphasis is being given to green growth, climate resilient infrastructure, and green transition.
Romanized Hindi: Bharat mein bhi green growth, climate resilient infrastructure aur green transition par vishesh roop se bal diya ja raha hai.
Devanagari Hindi: भारत में भी हरित विकास, जलवायु सहनशील आधारिक संरचना और हरित संक्रमण पर विशेष रूप से बल दिया जा रहा है।

---

6. Text Normalization (TN)

Sentence-level normalization of noisy Hindi-English code-mixed text.

Generated using GPT-OSS-120B, then refined manually.

Sentence: Janmdin ki dheron shubhkaamnaayen sir... Aap swasth rahen... Dirghayu ho... Yahi hamare poore pariwaar ki shubhkaamnaayen hain  
Normalized: Janmadin ki dheron shubhkaamnayein sir. Aap swasth rahe. Dirghayu ho, yahi hamare pure parivaar ki shubhkaamnayein hain.

---

📂 Repository Structure

.
├── LID_train.csv
├── LID_test.csv
├── MLI_train.csv
├── MLI_test.csv
├── MT_train.csv
├── MT_test.csv
├── NER_train.csv
├── NER_test.csv
├── POS_train.csv
├── POS_test.csv
├── TN_train.csv
└── TN_test.csv

---

🔗 Relevant Links

Paper Link: EMNLP 2025 Paper

Explore the Project: Project Website

📄 License

This dataset is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

---

⭐ Citation

If you find COMI-LINGUA helpful for your research, please cite it below:

@inproceedings{sheth-etal-2025-comi,
    title = "{COMI}-{LINGUA}: Expert Annotated Large-Scale Dataset for Multitask {NLP} in {H}indi-{E}nglish Code-Mixing",
    author = "Sheth, Rajvee  and
      Beniwal, Himanshu  and
      Singh, Mayank",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.findings-emnlp.422/",
    pages = "7973--7992",
    ISBN = "979-8-89176-335-7",
}