---
title: "TeamUNCC at SemEval-2018 Task 1: Emotion Detection in"
description: "The paper describes TeamUNCC's entry to SemEval-2018 Task 1 (Affect in Tweets). 0.23 Spearman. 29 references, cited by 11 papers here."
authors: ["Malak Abdullah", "Samira Shaikh"]
year: 2018
venue: "Proceedings of the 12th International Workshop on Semantic Evaluation"
source: "https://aclanthology.org/S18-1053/"
doi: "10.18653/v1/s18-1053"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/abdullah-2018-teamuncc-semeval-2018-task-1/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/abdullah-2018-teamuncc-semeval-2018-task-1/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2018](https://md.research.elamur.ai/free/en/papers/2018/page/4/) / TeamUNCC at SemEval-2018 Task 1: Emotion Detection in English and Arabic Tweets using Deep Learning

# TeamUNCC at SemEval-2018 Task 1: Emotion Detection in English and Arabic Tweets using Deep Learning

**Authors:** [Malak Abdullah](https://md.research.elamur.ai/free/en/authors/malak-abdullah/) and [Samira Shaikh](https://md.research.elamur.ai/free/en/authors/samira-shaikh/) · **Year:** 2018 · **Venue:** [Proceedings of the 12th International Workshop on Semantic Evaluation](https://md.research.elamur.ai/free/en/venues/semeval/) ([SemEval 2018](https://md.research.elamur.ai/free/en/venues/semeval/2018/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/S18-1053/) · [PDF](https://aclanthology.org/S18-1053.pdf) · [DOI 10.18653/v1/s18-1053](https://doi.org/10.18653/v1/s18-1053) · [OpenAlex](https://openalex.org/W2806752996)

**Citations:** 57 citations (OpenAlex, 2026-10-14) · [11](https://md.research.elamur.ai/free/en/papers/abdullah-2018-teamuncc-semeval-2018-task-1/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper describes TeamUNCC's entry to SemEval-2018 Task 1 (Affect in Tweets). The task asks systems to estimate emotion and sentiment intensity in tweets in five forms: EI-reg, EI-oc, V-reg, V-oc and E-c. The team took part in every subtask for both English and Arabic, and they report being the only team to do so. Their main design choice is a single shared pipeline for all subtasks and both languages.

Each tweet is encoded in several ways: lexicon and sentiment features from the AffectiveTweets Weka-package, doc2vec vectors, averaged word2vec vectors, and a zero-padded sequence of word2vec vectors. Arabic tweets are also machine-translated into English and processed as a second input. Three neural models are trained, two fully connected networks and one LSTM, and their scores are averaged into one intensity value. For the ordinal subtasks, that value is then cut into classes using fixed thresholds.

The system beats the organizers' baselines by 0.03 to 0.23 Spearman. It placed third in Arabic EI-oc and fourth in the other Arabic subtasks, using no external data. This shows that one simple, reusable architecture can stay competitive across languages and label formats.

## Key points
- Feature sets:
  - AffectTweets-145: 43 lexicon features, 2 sentiment-strength values and a 100-d embedding.
  - Doc2Vec-300: two 150-d doc2vec vectors.
  - Word2Vec-300: GoogleNews vectors for English, Twt-SG for Arabic.
  - PaddingWord2Vec-300: a per-word matrix padded to the longest tweet.
- Dense-Network: three ReLU layers of 256, 256 and 80 units, dropout of 0.3 and 0.5, and one sigmoid output. It is trained with SGD (lr 0.01, momentum 0.9) on MSE loss with early stopping. LSTM-Network: 256 LSTM units followed by a similar dense head.
- For the four intensity subtasks, three predictions are averaged. Arabic inputs combine the original tweets with English translations made by the python package translate 3.5.0.
- E-c is solved with one Dense-Network per emotion on Word2Vec-300, eleven in total, each with a 0.5 cutoff.
- Results (English / Arabic):
  - EI-reg average Spearman: 0.695 / 0.597.
  - EI-oc: 0.599 / 0.517.
  - V-reg: 0.787 / 0.773.
  - V-oc: 0.736 / 0.748.
  - E-c accuracy: 0.471 / 0.446.

## Interesting ideas
- The ordinal subtasks are handled by one continuous regressor plus per-emotion cut points. The cut points come from comparing tweets that carry both a real-valued and an ordinal label, which avoids training a separate classifier.
- Machine-translating Arabic tweets into English adds a second input view, letting English resources help a lower-resource language.
- The final score averages models that see the tweet through different representations: a lexicon/doc vector, an averaged embedding and an ordered sequence for the LSTM. Combining diverse views tends to reduce errors that all the models share.
- Hand-built lexicon scores are concatenated with learned embeddings, keeping some human-readable signal next to the opaque vectors.
- Multi-label emotion tagging is broken into independent yes/no decisions, one per emotion, which naturally lets several emotions co-occur.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (frames): prag could output a single graded score for sentiment or irony strength, then map it to discrete bands calibrated on doubly-annotated examples, so each band choice stays traceable to a threshold.
- **I2** (learn): paraphrased or back-translated versions of a sentence could be parsed as extra witnesses, and silver trees would be accepted only when the readings agree across views.
- **I3** (learn): the witness-parser pool should deliberately mix input representations, not only model seeds, so that agreement among witnesses is more meaningful.
- **I4** (world): affect-lexicon values could be stored as explicit attributes on entity states, attached to specific tree nodes rather than pooled, so emotional state remains explainable.
- **I5** (frames): speech-act or attitude labels in prag could be decided per label, so an utterance can be both a complaint and a request without a forced single choice.

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

## Institutions

- [University of North Carolina at Charlotte](https://md.research.elamur.ai/free/en/universities/university-of-north-carolina-at-charlotte/) — University, Charlotte, United States · 6 works · 125 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,552 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 870 works · [💭 Social Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/social-psychology/) · 306 works

🧭 Topics (3): [English](https://md.research.elamur.ai/free/en/topics/english/) · 2,263 works · [😊 Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · 972 works · [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

## Cite

Cite as: Malak Abdullah and Samira Shaikh. 2018. TeamUNCC at SemEval-2018 Task 1: Emotion Detection in English and Arabic Tweets using Deep Learning. Proceedings of the 12th International Workshop on Semantic Evaluation.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{abdullah2018teamuncc,
  title = {{TeamUNCC at SemEval-2018 Task 1: Emotion Detection in English and Arabic Tweets using Deep Learning}},
  author = {Malak Abdullah and Samira Shaikh},
  booktitle = {Proceedings of the 12th International Workshop on Semantic Evaluation},
  year = {2018},
  doi = {10.18653/v1/s18-1053},
  url = {https://doi.org/10.18653/v1/s18-1053}
}
```

- [← previous Improving a Neural Semantic Parser by Counterfactual Learning from Human Bandit Feedback](https://md.research.elamur.ai/free/en/papers/lawrence-2018-improving-neural-semantic-parser-counterfactual/)
- [next → Deep Reinforcement Learning for Chinese Zero Pronoun Resolution](https://md.research.elamur.ai/free/en/papers/yin-2018-deep-reinforcement-learning-chinese-zero/)
- [↑ back to papers of 2018](https://md.research.elamur.ai/free/en/papers/2018/page/4/)

## References

29 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Malak Abdullah and Mirsad Hadzikadic. 2017. Sentiment analysis on arabic tweets: Challenges to dissecting the language. In International Conference on Social Computing and Social Media, pages 191– 202. Springer. [find](https://scholar.google.com/scholar?q=Malak+Abdullah+and+Mirsad+Hadzikadic%2E+2017%2E+Sentiment+analysis+on+arabic+tweets%3A+Challenges+to+dissecting+the+language%2E+In+International+Conference+on+Social+Computing+and+Social+Media%2C+pages+191%E2%80%93+202)
2. Apoorv Agarwal, Boyi Xie, Ilia Vovsha, Owen Rambow, and Rebecca Passonneau. 2011. [Sentiment analysis of twitter data](https://md.research.elamur.ai/free/en/papers/agarwal-2011-sentiment-analysis-twitter-data/). In Proceedings of the workshop on languages in social media, pages 30–38. Association for Computational Linguistics.
3. Adel Assiri, Ahmed Emam, and Hmood Al-Dossari. 2016. Saudi twitter corpus for sentiment analysis. World Academy of Science, Engineering and Technology, International Journal of Computer, Electrical, Automation, Control and Information Engineering, 10(2):272–275. [↗](https://doi.org/10.5281/zenodo.1338815)
4. Adel Assiri, Ahmed Emam, and Hmood Aldossari. 2015. Arabic sentiment analysis: a survey. [↗](https://doi.org/10.14569/ijacsa.2015.061211)
5. Interna-Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, and Jeff Dean. 2013b. [Distributed representations of words and phrases and their compositionality](https://md.research.elamur.ai/free/en/papers/mikolov-2013-distributed-representations-words-phrases-their/). In Advances in neural information processing systems, pages 3111–3119. tional Journal of Advanced Computer Science and Applications, 6(12):75–85. [↗](https://doi.org/10.48550/arxiv.1310.4546)
6. Felipe Bravo-Marquez, Marcelo Mendoza, and Barbara Poblete. 2014. Meta-level sentiment models for big social data analysis. Knowledge-Based Systems, 69:86–99. [↗](https://doi.org/10.1016/j.knosys.2014.05.016)
7. Saif M Mohammad. 2012. [# emotional tweets](https://md.research.elamur.ai/free/en/papers/mohammad-2012-emotional-tweets/). In Proceedings of the First Joint Conference on Lexical and Computational Semantics-Volume 1: Proceedings of the main conference and the shared task, and Volume 2: Proceedings of the Sixth International Workshop on Semantic Evaluation, pages 246–255. Association for Computational Linguistics.
8. Delenn Chin, Anna Zappone, and Jessica Zhao. 2016. Analyzing twitter sentiment of the 2016 presidential candidates. News & Publications: Stanford University. [find](https://scholar.google.com/scholar?q=Delenn+Chin%2C+Anna+Zappone%2C+and+Jessica+Zhao%2E+2016%2E+Analyzing+twitter+sentiment+of+the+2016+presidential+candidates%2E+News+%26+Publications%3A+Stanford+University%2E)
9. Paul Ekman. 1993. Facial expression and emotion. American psychologist, 48(4):384. [↗](https://doi.org/10.1037//0003-066x.48.4.384)
10. Saif M Mohammad and Felipe Bravo-Marquez. 2017. [Wassa-2017 shared task on emotion intensity](https://md.research.elamur.ai/free/en/papers/mohammad-2017-wassa-2017-shared-task-emotion/). arXiv preprint arXiv:1708.03700. [↗](https://arxiv.org/abs/1708.03700)
11. Samhaa R El-Beltagy, Talaat Khalil, Amal Halaby, and Muhammad Hammad. 2017. Combining lexical features and a supervised learning approach for arabic sentiment analysis. arXiv preprint arXiv:1710.08451. [↗](https://arxiv.org/abs/1710.08451)
12. Saif M. Mohammad, Felipe Bravo-Marquez, Mohammad Salameh, and Svetlana Kiritchenko. 2018. [Semeval-2018 Task 1: Affect in tweets](https://md.research.elamur.ai/free/en/papers/mohammad-2018-semeval-2018-task-1-affect/). In Proceedings of International Workshop on Semantic Evaluation (SemEval-2018), New Orleans, LA, USA.
13. Pranav Goel, Devang Kulshreshtha, Prayas Jain, and Kaushal Kumar Shukla. 2017. Prayas at emoint 2017: An ensemble of deep neural architectures for emotion intensity prediction in tweets. In Proceedings of the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis, pages 58–65. [find](https://scholar.google.com/scholar?q=Pranav+Goel%2C+Devang+Kulshreshtha%2C+Prayas+Jain%2C+and+Kaushal+Kumar+Shukla%2E+2017%2E+Prayas+at+emoint+2017%3A+An+ensemble+of+deep+neural+architectures+for+emotion+intensity+prediction+in+tweets%2E+In+Proceeding)
14. Saif M Mohammad and Svetlana Kiritchenko. 2015. Using hashtags to capture fine emotion categories from tweets. Computational Intelligence, 31(2):301–326. [↗](https://doi.org/10.1111/coin.12024)
15. Saif M. Mohammad and Svetlana Kiritchenko. 2018. Understanding emotions: A dataset of tweets to study interactions between affect categories. In Proceedings of the 11th Edition of the Language Resources and Evaluation Conference, Miyazaki, Japan. [find](https://scholar.google.com/scholar?q=Saif+M%2E+Mohammad+and+Svetlana+Kiritchenko%2E+2018%2E+Understanding+emotions%3A+A+dataset+of+tweets+to+study+interactions+between+affect+categories%2E+In+Proceedings+of+the+11th+Edition+of+the+Language+Resourc)
16. Salud Maria Jimenez-Zafra, M Teresa Martin Valdivia, Eugenio Martinez Camara, and Luis Alfonso Urena-Lopez. 2017. Studying the scope of negation for spanish sentiment analysis on twitter. IEEE Transactions on Affective Computing. [↗](https://doi.org/10.1109/taffc.2017.2693968)
17. Svetlana Kiritchenko, Saif Mohammad, and Mohammad Salameh. 2016. [Semeval-2016 task 7: Determining sentiment intensity of english and arabic phrases](https://md.research.elamur.ai/free/en/papers/kiritchenko-2016-semeval-2016-task-7-determining/). In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval-2016), pages 42–51.
18. Tetsuya Nasukawa and Jeonghee Yi. 2003. Sentiment analysis: Capturing favorability using natural language processing. In Proceedings of the 2nd international conference on Knowledge capture, pages 70–77. ACM. [find](https://scholar.google.com/scholar?q=Tetsuya+Nasukawa+and+Jeonghee+Yi%2E+2003%2E+Sentiment+analysis%3A+Capturing+favorability+using+natural+language+processing%2E+In+Proceedings+of+the+2nd+international+conference+on+Knowledge+capture%2C+pages+70%E2%80%93)
19. Efthymios Kouloumpis, Theresa Wilson, and Johanna D Moore. 2011. Twitter sentiment analysis: The good the bad and the omg! Icwsm, 11(538-541):164. [find](https://scholar.google.com/scholar?q=Efthymios+Kouloumpis%2C+Theresa+Wilson%2C+and+Johanna+D+Moore%2E+2011%2E+Twitter+sentiment+analysis%3A+The+good+the+bad+and+the+omg%21+Icwsm%2C+11%28538%2D541%29%3A164%2E)
20. Nels Oscar, Pamela A Fox, Racheal Croucher, Riana Wernick, Jessica Keune, and Karen Hooker. 2017. Machine learning, sentiment analysis, and tweets: an examination of alzheimers disease stigma on twitter. Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 72(5):742–751. [↗](https://doi.org/10.1093/geronb/gbx014)
21. Jey Han Lau and Timothy Baldwin. 2016. An empirical evaluation of doc2vec with practical insights into document embedding generation. arXiv preprint arXiv:1607.05368. [↗](https://arxiv.org/abs/1607.05368)
22. Alexander Pak and Patrick Paroubek. 2010. [Twitter as a corpus for sentiment analysis and opinion mining](https://md.research.elamur.ai/free/en/papers/pak-2010-twitter-as-corpus-sentiment-analysis/). In LREc, volume 10.
23. Quoc Le and Tomas Mikolov. 2014. [Distributed representations of sentences and documents](https://md.research.elamur.ai/free/en/papers/le-2014-distributed-representations-sentences-documents/). In International Conference on Machine Learning, pages 1188–1196. [↗](https://doi.org/10.48550/arxiv.1405.4053)
24. Omneya Rabie and Christian Sturm. 2014. Feel the heat: Emotion detection in arabic social media content. In The International Conference on Data Mining, Internet Computing, and Big Data (Big-Data2014), pages 37–49. The Society of Digital Information and Wireless Communication. [find](https://scholar.google.com/scholar?q=Omneya+Rabie+and+Christian+Sturm%2E+2014%2E+Feel+the+heat%3A+Emotion+detection+in+arabic+social+media+content%2E+In+The+International+Conference+on+Data+Mining%2C+Internet+Computing%2C+and+Big+Data+%28Big%2DData2014%29)
25. Andrew L Maas, Awni Y Hannun, and Andrew Y Ng. 2013. Rectifier nonlinearities improve neural network acoustic models. In Proc. icml, volume 30, page 3. [find](https://scholar.google.com/scholar?q=Andrew+L+Maas%2C+Awni+Y+Hannun%2C+and+Andrew+Y+Ng%2E+2013%2E+Rectifier+nonlinearities+improve+neural+network+acoustic+models%2E+In+Proc%2E+icml%2C+volume+30%2C+page+3%2E)
26. Sara Rosenthal, Noura Farra, and Preslav Nakov. 2017. [Semeval-2017 task 4: Sentiment analysis in twitter](https://md.research.elamur.ai/free/en/papers/rosenthal-2017-semeval-2017-task-4-sentiment/). In Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pages 502–518.
27. Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013a. [Efficient estimation of word representations in vector space](https://md.research.elamur.ai/free/en/papers/mikolov-2013-efficient-estimation-word-representations-vector/). arXiv preprint arXiv:1301.3781. [↗](https://arxiv.org/abs/1301.3781)
28. Abu Bakr Soliman, Kareem Eissa, and Samhaa R El-Beltagy. 2017. Aravec: A set of arabic word embedding models for use in arabic nlp. Procedia Computer Science, 117:256–265. [find](https://scholar.google.com/scholar?q=Abu+Bakr+Soliman%2C+Kareem+Eissa%2C+and+Samhaa+R+El%2DBeltagy%2E+2017%2E+Aravec%3A+A+set+of+arabic+word+embedding+models+for+use+in+arabic+nlp%2E+Procedia+Computer+Science%2C+117%3A256%E2%80%93265%2E)
29. Erik Tromp and Mykola Pechenizkiy. 2014. Rule-based emotion detection on social media: putting tweets on plutchik’s wheel. arXiv preprint arXiv:1412.4682. [↗](https://arxiv.org/abs/1412.4682)

*Updated 1 November 2026*