Elamur Research⌕
☰
home / papers / 2026 / SJM_MINDS@DravidianLangTech@ACL2026: Machine Learning Approaches for Hope Speech Detection in Code-Mixed Tulu

SJM_MINDS@DravidianLangTech@ACL2026: Machine Learning Approaches for Hope Speech Detection in Code-Mixed Tulu

Paper · 2026 · Hosahalli Lakshmaiah Shashirekha, Manjula and Jayashree Krishna

Summary

The paper addresses hope speech detection in code-mixed Tulu, a low-resource Dravidian language whose social media comments mix Tulu with Kannada and English. They also show spelling and transliteration inconsistencies. The authors, from Mangalore University, entered the first shared task on this problem at DravidianLangTech@ACL 2026. The task had two subtasks:

  • Task 1: a four-way classification of hope tone.
  • Task 2: a five-way classification of hope type.

Their system is a classical pipeline: lower-casing, punctuation removal and tokenization, followed by TF-IDF features over word unigrams and bigrams. These features feed two linear classifiers, Logistic Regression and LinearSVC. Both models were tried for Task 1, but only LR was submitted for Task 2. On the test sets, LinearSVC placed 4th in Task 1 with a macro F1 of 0.51, and LR placed 5th in Task 2 with 0.37.

Error analysis shows recurring confusions:

  • Mixed or negative tones are often predicted as neutral or as other classes.
  • Positive hope subtypes tend to collapse into Inspiring Hope.

The work matters because it offers a lightweight, reproducible baseline for an underexplored language. It suggests that inexpensive linear models stay competitive where pretrained resources are scarce.

Key points

  • Shared task data:
    • Task 1 (Encouraging, Discouraging, Uninvolved, Blended Tone) has 5,991 training, 1,284 development and 1,284 test comments.
    • Task 2 (Optimistic, Realistic, Inspiring, Fading Hope, Hopelessness) has 3,185 training, 682 development and 682 test comments. The task is hosted on Codabench.
  • Model settings:
    • LR uses at most 5,000 TF-IDF features and 1,000 iterations.
    • LinearSVC uses 8,000 features, min_df = 2 and class_weight = balanced to counter class imbalance.
  • Test results:
    • LinearSVC on Task 1 reached precision 0.52, recall 0.51, macro F1 0.51 and accuracy 0.62.
    • LR on Task 2 reached precision 0.64, recall 0.39, macro F1 0.37 and accuracy 0.56.
  • Hardest classes: Fading Hope was the weakest category, with very few correct predictions. Discouraging and Blended Tone were also frequently missed.
  • Stated limitations: the dataset is small, the classes are imbalanced, the text is noisy code-mixed Tulu, and TF-IDF cannot capture deeper semantics. The authors plan more feature engineering and further ML models as future work.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksPsychology · 1,552 worksExperimental and Cognitive Psychology · 870 worksSocial Psychology · 306 works

🧭 Topics 1Research subjects and languages the work is about.Multilingual & cross-lingual NLP · 828 works

Cite

Cite as Hosahalli Lakshmaiah Shashirekha, Manjula and Jayashree Krishna. 2026. SJM_MINDS@DravidianLangTech@ACL2026: Machine Learning Approaches for Hope Speech Detection in Code-Mixed Tulu. Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.dravidianlangtech-1.56

@inproceedings{shashirekha2026sjm,
  title = {{SJM\_MINDS@DravidianLangTech@ACL2026: Machine Learning Approaches for Hope Speech Detection in Code-Mixed Tulu}},
  author = {Hosahalli Lakshmaiah Shashirekha, Manjula and Jayashree Krishna},
  booktitle = {Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages},
  year = {2026},
  doi = {10.18653/v1/2026.dravidianlangtech-1.56},
  url = {https://doi.org/10.18653/v1/2026.dravidianlangtech-1.56}
}

References

17 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Oluwatobi Joseph Abiola, Tolulope Olalekan Abiola, Obdulia Pichardo Lagunas, and Grigori Sidorov. 2026. Hope Speech Detection: A Comparative Study across Four Languages. International Journal of Combinatorial Optimization Problems and Informatics, 17(2):215–230. find
  2. Pranjal Aggarwal, Pasupuleti Chandana, Jagrut Nemade, Shubham Sharma, Sunil Saumya, and Shankar Biradar. 2023. Hope Speech Detection on Social Media Platforms. In Cybercrime in Social Media, pages 67–84. Chapman and Hall/CRC. find
  3. Fazlourrahman Balouchzahi, Grigori Sidorov, and Alexander Gelbukh. 2023. Polyhope: Two-Level Hope Speech Detection from Tweets. Expert Systems with Applications, 225:120078. find
  4. Bharathi Raja Chakravarthi, Prasanna Kumar Kumaresan, Ruba Priyadharshini, Paul Buitelaar, Asha Hegde, Hosahalli Lakshmaiah Shashirekha, Saranya Rajiakodi, Miguel Ángel García-Cumbreras, Salud María Jiménez-Zafra, José Antonio García-Díaz, Rafael Valencia-García, Kishore Kumar Ponnusamy, Poorvi Shet… find
  5. Divakaran, K. Girish, and H. L. Shashirekha. 2024. Hope on the Horizon: Experiments with Learning Models for Hope Speech Detection in Spanish and English. In Proceedings of the Iberian Languages Evaluation Forum (IberLEF 2024), Co-located with the 40th Conference of the Spanish Society for Natural L… find
  6. Asha Hegde, Bharathi Raja Chakravarthi, Hosahalli Lakshmaiah Shashirekha, Rahul Ponnusamy, Subalalitha Chinnaudayar Navaneethakrishnan, Lavanya Sambath Kumar, Durairaj Thenmozhi, Martha Karunakar, Shreya Sriram, and Sarah Aymen. 2023a. Findings of the Shared Task on Sentiment Analysis in Tamil and T… find
  7. Asha Hegde, Kavya G, Sharal Coelho, and Hosahalli Lakshmaiah Shashirekha. 2023b. MUCS@LT-EDI2023: Learning Approaches for Hope Speech Detection in Social Media Text. In Proceedings of the Third Workshop on Language Technology for Equality, Diversity and Inclusion, pages 279–286, Varna, Bulgaria. INC… find
  8. Asha Hegde and Hosahalli Lakshmaiah Shashirekha. 2023. KT2: Kannada–Tulu Parallel Corpus Construction for Neural Machine Translation. In Proceedings of the 20th International Conference on Natural Language Processing (ICON), pages 743– 753, Goa University, Goa, India. NLP Association of India (NLPAI… find
  9. Asha Hegde and Hosahalli Lakshmaiah Shashirekha. 2024. Syllable-Level Morphological Segmentation of Kannada and Tulu Words, pages 113–133. Wiley, Hoboken, NJ, USA. find
  10. Prasanna Kumar Kumaresan, Bharathi Raja Chakravarthi, Subalalitha Cn, Miguel Ángel García-Cumbreras, Salud María Jiménez Zafra, José Antonio García-Díaz, Rafael Valencia-García, Momchil Hardalov, Ivan Koychev, Preslav Nakov, Daniel García-Baena, and Kishore Kumar Ponnusamy. 2023. Overview of the Sha… find
  11. GodsGift Uzor, Mohammadsaleh Pajuhanfard, Michael Beebe, Maaz Amjad, and Victor Sheng. 2025. Hope Speech Detection: A Comparative Study of Machine Learning, Deep Learning, and Transformer-Based Models. Journal of Artificial Intelligence Research, 74:123–145. find
  12. Mesay Gemeda Yigezu, Girma Yohannis Bade, Olga Kolesnikova, Grigori Sidorov, and Alexander F Gelbukh. 2023. Multilingual Hope Speech Detection Using Machine Learning. In Proceedings of the Iber-LEF 2023 Workshop at SEPLN, volume 3498, pages 123–134. CEUR Workshop Proceedings. find
  13. Manu Narayanan and Noëmi Aepli. 2024. A Tulu Resource for Machine Translation. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 1756–1767, Torino, Italia. ELRA and ICCL. find
  14. Luis Ramos, Hiram Calvo, and Olga Kolesnikova. 2025. Hope Speech Detection in Social Media English Corpora: Performance of Traditional and Transformer Models. Preprint, arXiv:2510.23585.
  15. Deepawali Sharma, Vedika Gupta, Vivek Kumar Singh, and Bharathi Raja Chakravarthi. 2025. Stop the Hate, Spread the Hope: An Ensemble Model for Hope Speech Detection in English and Dravidian Languages. ACM Transactions on Asian and Low-Resource Language Information Processing, 24(3):1– 18. find
  16. Durairaj Thenmozhi, Bharathi Raja Chakravarthi, Asha Hegde, Hosahalli Lakshmaiah Shashirekha, Rajeswari Natarajan, Sajeetha Thavareesan, Ratnasingam Sakuntharaj, Krishnakumari Kalyanasun-daram, Charmathi Rajkumar, Poorvi Shetty, and Harshitha S Kumar. 2025. Overview of the Shared Task on Sentiment A… find
  17. Durairaj Thenmozhi, Rathnakar Shetty P, Parameshwar R. Hegde, M D Anusha, Raksha Adyanthaya, Mohammed Fadhel Aljunid, Prasanna Kumar Kumaresan, and Bharathi Raja Chakravarthi. 2026. Findings of the Shared Task on Hope Speech Detection in Tulu. In Proceedings of the Sixth Workshop on Speech, Vision,… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/shashirekha-2026-sjm-minds-dravidianlangtech-acl2026-machine/