Code-mixing, the dynamic interplay of multiple languages within a single discourse, is a widespread linguistic phenomenon observed in multilingual societies. Code-mixing is particularly intriguing when observed in closely related languages.


We invite you to participate in our shared task at the WILDRE workshop, which is co-located with LREC-COLING 2024. This shared task addresses the complexities of code-mixed data from less-resourced similar languages for sentiment analysis. We will provide annotated data for the following code-mixed languages:

1. Magahi-Hindi-English
2. Bangla-English-Hindi
3. Hindi-English

The evaluation will be in two different Tracks:

*A. Track 1:* Given training and validation data to determine the comment's polarity (positive, negative, neutral or mixed) in the same code-mixed setting.

1. Hindi-English
2. Magahi-Hindi-English
3. Bangla-English
4. Combined all the language pairs (1+2+3)

*B. Track 2:* Given unlabelled test data for the code-mixed Maithili language (Maithi-Hindi-English), leverage any or all of the available training datasets in Track 1 to determine the sentiment of a comment in the target language.

Important Links:
Important Dates:
  • Dec 22, 2023: Registration
  • Jan 10, 2024: Train and Validation Data set Release [to get the data, please register]
  • Feb 15, 2024: Test Set Release
  • Feb 23, 2024: System Submission Due
  • Feb 29, 2024: System Results
  • March 15, 2024: System Description Paper Due
  • March 28, 2024: Paper notification of acceptance