---
title: "UNIBA — paper by Basile and Novielli (2015)"
description: "The paper describes UNIBA's entry to SemEval-2015 Task 10, subtask B. The task is to… 3.29 points. 19 references, cited by 1 paper here."
authors: ["Pierpaolo Basile", "Nicole Novielli"]
year: 2015
venue: "Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015)"
source: "https://aclanthology.org/S15-2099/"
doi: "10.18653/v1/s15-2099"
licence: "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)"
licence_url: "https://creativecommons.org/licenses/by-nc-sa/3.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/basile-2015-uniba-sentiment-analysis-english-tweets/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/basile-2015-uniba-sentiment-analysis-english-tweets/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2015](https://md.research.elamur.ai/free/en/papers/2015/page/3/) / UNIBA: Sentiment Analysis of English Tweets Combining Micro-blogging, Lexicon and Semantic Features

# UNIBA: Sentiment Analysis of English Tweets Combining Micro-blogging, Lexicon and Semantic Features

**Authors:** [Pierpaolo Basile](https://md.research.elamur.ai/free/en/authors/pierpaolo-basile/) and [Nicole Novielli](https://md.research.elamur.ai/free/en/authors/nicole-novielli/) · **Year:** 2015 · **Venue:** [Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015)](https://md.research.elamur.ai/free/en/venues/semeval/) ([SemEval 2015](https://md.research.elamur.ai/free/en/venues/semeval/2015/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY-NC-SA 3.0 (ACL Anthology, before 2016)](https://creativecommons.org/licenses/by-nc-sa/3.0/)

**Links:** [Publisher's page](https://aclanthology.org/S15-2099/) · [PDF](https://aclanthology.org/S15-2099.pdf) · [DOI 10.18653/v1/s15-2099](https://doi.org/10.18653/v1/s15-2099) · [OpenAlex](https://openalex.org/W2167307309)

**Citations:** 13 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper describes UNIBA's entry to SemEval-2015 Task 10, subtask B. The task is to label each English tweet as positive, negative or neutral. The authors adapted their Italian tweet classifier, which won Evalita 2014, and trained a linear SVM (Liblinear, L2-loss) on the 8,006 provided training tweets. The model uses three feature families:
- **Surface and micro-blogging cues:** n-grams marked for negation, metatokens, emoticons, character repetition and Twitter word clusters.
- **Lexicon scores:** prior-polarity counts from eight sentiment and psycholinguistic lexicons, computed separately for affirmative and negated spans.
- **Distributional semantics:** a word2vec space built from 15M self-collected tweets, in which each tweet vector is compared to lexicon-derived class prototypes.

On the official 2015 test set the system scored 61.55 averaged positive/negative F and ranked 12th of 40, 3.29 points behind the winner. It held up on SMS and LiveJournal data but dropped sharply on sarcastic tweets. Ablation showed the lexicon features matter most, which differs from the authors' Italian results. This makes the paper a useful case study of how feature value shifts across languages and resources.

## Key points
- **Data:** 8,006 training tweets, a 2,390-tweet official test set and an 8,987-tweet progress set. Additional development data was not used for training.
- **Results:**
  - Official 2015 set: 61.55, rank 12/40.
  - Twitter 2014: 65.11, rank 25/40.
  - LiveJournal 2014: 70.05.
  - SMS 2013: 65.50.
  - Twitter 2014 Sarcasm: only 37.30.
- **Ablation (three-class F, baseline 64.95):**
  - Without lexicons: 60.93 (−6.19%).
  - Without keyword features: 63.41 (−2.37%).
  - Without semantic features: 64.50 (−0.69%).
- **Lexicons used:** NRC Emotion, MPQA, Bing Liu, NRC Hashtag Sentiment, Sentiment140, SentiStrength, a Wikipedia emoticon list, SentiWordNet 3.0 (most frequent sense per POS), and LIWC word classes.
- **Setup:**
  - word2vec skip-gram with 300 dimensions and a minimum count of 10, giving 308,493 terms.
  - CMU TweetNLP tokenizer and POS tagger.
  - C=0.01, chosen by 10-fold cross-validation.
  - 145,967 features in total.

## Interesting ideas
- Every lexicon statistic is computed twice, once for words inside a negation span and once for words outside it, so the classifier learns how negation shifts each cue.
- Class "prototypes" are built by summing the vectors of all lexicon words of one polarity, and an input is then scored by its cosine similarity to each prototype, which gives a cheap and inspectable link between a lexicon and an embedding space.
- The unlabeled corpus for the embeddings was collected by querying the stream with class-discriminative words chosen by KL divergence, with equal amounts per class, so the space is biased toward relevant vocabulary.
- Ablating whole feature groups and comparing the outcome with an earlier study in another language showed that feature usefulness is not portable, for example semantic features mattered greatly in Italian but barely in English.
- The authors traced many errors to irony and sarcasm needing common-sense reasoning, kept a dedicated sarcasm test split, and treated the large drop there as a diagnostic rather than averaging it away.

## For Mova

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

- **I1** (frames): compute negation scope from the UD tree, since `neg`/`advmod` heads give clean scope boundaries. Polarity and stance in frames could then carry an explicit "under negation" flag instead of flipping a sign blindly.
- **I2** (world): a frame argument could be compared against prototypes built from WordNet or VerbNet class members, as an explainable soft check on class membership that reports which prototype it was closest to.
- **I3** (learn): when gathering raw text for silver parsing, seed the collection with KLD-selected terms for under-covered phenomena, such as rare constructions or domain verbs, and balance the buckets.
- **I4** (learn): ablate entire rule or feature families per corpus and genre as standard evaluation, and record when a component's contribution flips between datasets.
- **I5** (frames): keep a separate irony and sarcasm evaluation slice for the implicature and irony module, so that gains on literal text cannot hide a failure on non-literal text.

*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 Bari Aldo Moro](https://md.research.elamur.ai/free/en/universities/university-of-bari-aldo-moro/) — University, Bari, Italy · 6 works · 2,210 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,649 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 930 works

🧭 Topics (2): [English](https://md.research.elamur.ai/free/en/topics/english/) · 2,379 works · [😊 Sentiment & emotion](https://md.research.elamur.ai/free/en/topics/sentiment-emotion/) · 1,035 works

## Cite

Cite as: Pierpaolo Basile and Nicole Novielli. 2015. UNIBA: Sentiment Analysis of English Tweets Combining Micro-blogging, Lexicon and Semantic Features. Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015).

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

```bibtex
@inproceedings{basile2015uniba,
  title = {{UNIBA: Sentiment Analysis of English Tweets Combining Micro-blogging, Lexicon and Semantic Features}},
  author = {Pierpaolo Basile and Nicole Novielli},
  booktitle = {Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015)},
  year = {2015},
  doi = {10.18653/v1/s15-2099},
  url = {https://doi.org/10.18653/v1/s15-2099}
}
```

- [← previous Combining Minimally-supervised Methods for Arabic Named Entity Recognition](https://md.research.elamur.ai/free/en/papers/althobaiti-2015-combining-minimally-supervised-methods-arabic/)
- [next → A Study of the Effect of Resolving Negation and Sentiment Analysis in Recognizing Text Entailment for Arabic](https://md.research.elamur.ai/free/en/papers/al-khawaldeh-2019-study-effect-resolving-negation-sentiment/)
- [↑ back to papers of 2015](https://md.research.elamur.ai/free/en/papers/2015/page/3/)

## References

19 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Pierpaolo Basile and Nicole Novielli. 2014. UNIBA at EVALITA 2014-SENTIPOLC Task: Predicting tweet sentiment polarity combining micro-blogging, lexicon and semantic features. In Proceedings of EVALITA 2014, pages 58–63. [find](https://scholar.google.com/scholar?q=Pierpaolo+Basile+and+Nicole+Novielli%2E+2014%2E+UNIBA+at+EVALITA+2014%2DSENTIPOLC+Task%3A+Predicting+tweet+sentiment+polarity+combining+micro%2Dblogging%2C+lexicon+and+semantic+features%2E+In+Proceedings+of+EVALITA)
2. Valerio Basile, Andrea Bolioli, Malvina Nissim, Viviana Patti, and Paolo Rosso. 2014. Overview of the Evalita 2014 SENTIment POLarity Classification Task. In Proc. of EVALITA 2014, Pisa, Italy. [find](https://scholar.google.com/scholar?q=Valerio+Basile%2C+Andrea+Bolioli%2C+Malvina+Nissim%2C+Viviana+Patti%2C+and+Paolo+Rosso%2E+2014%2E+Overview+of+the+Evalita+2014+SENTIment+POLarity+Classification+Task%2E+In+Proc%2E+of+EVALITA+2014%2C+Pisa%2C+Italy%2E)
3. Andrea Esuli and Fabrizio Sebastiani. 2006. [Sentiwordnet: A publicly available lexical resource for opinion mining](https://md.research.elamur.ai/free/en/papers/esuli-2006-sentiwordnet-publicly-available-lexical-resource/). In Proceedings of LREC, volume 6, pages 417–422.
4. Minqing Hu and Bing Liu. 2004. Mining and Summarizing Customer Reviews. In Proceedings of the Tenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 168–177. [find](https://scholar.google.com/scholar?q=Minqing+Hu+and+Bing+Liu%2E+2004%2E+Mining+and+Summarizing+Customer+Reviews%2E+In+Proceedings+of+the+Tenth+ACM+SIGKDD+International+Conference+on+Knowledge+Discovery+and+Data+Mining%2C+pages+168%E2%80%93177%2E)
5. Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. [Efficient Estimation of Word Representations in Vector Space](https://md.research.elamur.ai/free/en/papers/mikolov-2013-efficient-estimation-word-representations-vector/). CoRR, abs/1301.3781. [↗](https://arxiv.org/abs/1301.3781)
6. Yasuhide Miura, Shigeyuki Sakaki, Keigo Hattori, and Tomoko Ohkuma. 2014. TeamX: A Sentiment Analyzer with Enhanced Lexicon Mapping and Weighting Scheme for Unbalanced Data. In Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 628–632, Dublin, Ireland, August. [find](https://scholar.google.com/scholar?q=Yasuhide+Miura%2C+Shigeyuki+Sakaki%2C+Keigo+Hattori%2C+and+Tomoko+Ohkuma%2E+2014%2E+TeamX%3A+A+Sentiment+Analyzer+with+Enhanced+Lexicon+Mapping+and+Weighting+Scheme+for+Unbalanced+Data%2E+In+Proceedings+of+the+8th+)
7. Saif M. Mohammad and Peter D. Turney. 2010. [Emotions Evoked by Common Words and Phrases: Using Mechanical Turk to Create an Emotion Lexicon](https://md.research.elamur.ai/free/en/papers/mohammad-2010-emotions-evoked-common-words-phrases/). In Proceedings of the NAACL HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, CAAGET ’10, pages 26–34.
8. Saif Mohammad, Svetlana Kiritchenko, and Xiaodan Zhu. 2013. [Nrc-canada: Building the state-of-the-art in sentiment analysis of tweets](https://md.research.elamur.ai/free/en/papers/mohammad-2013-nrc-canada-building-state-art/). In Second Joint Conference on Lexical and Computational Semantics (*SEM), Volume 2: Proceedings of the Seventh International Workshop on Semantic Evaluation (SemEval 2013), pages 321–327, Atlanta, Georgia, USA, June.
9. Nicole Novielli and Carlo Strapparava. 2013. The Role of Affect Analysis in Dialogue Act Identification. IEEE Transactions on Affective Computing, 4:439– 451. [find](https://scholar.google.com/scholar?q=Nicole+Novielli+and+Carlo+Strapparava%2E+2013%2E+The+Role+of+Affect+Analysis+in+Dialogue+Act+Identification%2E+IEEE+Transactions+on+Affective+Computing%2C+4%3A439%E2%80%93+451%2E)
10. Bo Pang and Lillian Lee. 2008. Opinion Mining and Sentiment Analysis. Foundations and trends in information retrieval, 2(1-2):1–135, January. [find](https://scholar.google.com/scholar?q=Bo+Pang+and+Lillian+Lee%2E+2008%2E+Opinion+Mining+and+Sentiment+Analysis%2E+Foundations+and+trends+in+information+retrieval%2C+2%281%2D2%29%3A1%E2%80%93135%2C+January%2E)
11. Bo Pang, Lillian Lee, and Shivakumar Vaithyanathan. 2002. [Thumbs up? Sentiment Classification using Machine Learning Techniques](https://md.research.elamur.ai/free/en/papers/pang-2002-thumbs-up-sentiment-classification-machine/). In Proceedings of the 2002 Conference on Empirical Methods in Natural Language Processing, pages 79–86, July.
12. J. Pennebaker and M. Francis. 2001. Linguistic Inquiry and Word Count: LIWC. Erlbaum Publishers. [find](https://scholar.google.com/scholar?q=J%2E+Pennebaker+and+M%2E+Francis%2E+2001%2E+Linguistic+Inquiry+and+Word+Count%3A+LIWC%2E+Erlbaum+Publishers%2E)
13. Sara Rosenthal, Alan Ritter, Preslav Nakov, and Veselin Stoyanov. 2014. [SemEval-2014 Task 9: Sentiment Analysis in Twitter](https://md.research.elamur.ai/free/en/papers/rosenthal-2014-semeval-2014-task-9-sentiment/). In Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 73–80, Dublin, Ireland, August.
14. Sara Rosenthal, Preslav Nakov, Svetlana Kiritchenko, Saif M Mohammad, Alan Ritter, and Veselin Stoyanov. 2015. [SemEval-2015 Task 10: Sentiment Analysis in Twitter](https://md.research.elamur.ai/free/en/papers/rosenthal-2015-semeval-2015-task-10-sentiment/). In Proceedings of the 9th International Workshop on Semantic Evaluation, SemEval ’2015, Denver, Colorado, June.
15. Magnus Sahlgren. 2006. The Word-Space Model: Using distributional analysis to represent syntagmatic and paradigmatic relations between words in high-dimensional vector spaces. Ph.D. thesis. [find](https://scholar.google.com/scholar?q=Magnus+Sahlgren%2E+2006%2E+The+Word%2DSpace+Model%3A+Using+distributional+analysis+to+represent+syntagmatic+and+paradigmatic+relations+between+words+in+high%2Ddimensional+vector+spaces%2E+Ph%2ED%2E+thesis%2E)
16. Paul Smolensky. 1990. Tensor product variable binding and the representation of symbolic structures in connectionist systems. Artificial Intelligence, 46(1-2):159–216, November. [find](https://scholar.google.com/scholar?q=Paul+Smolensky%2E+1990%2E+Tensor+product+variable+binding+and+the+representation+of+symbolic+structures+in+connectionist+systems%2E+Artificial+Intelligence%2C+46%281%2D2%29%3A159%E2%80%93216%2C+November%2E)
17. Mike Thelwall, Kevan Buckley, Georgios Paltoglou, Di Cai, and Arvid Kappas. 2010. Sentiment in Short Strength Detection Informal Text. Journal of the American Society for Information Science and Technology, 61(12):2544–2558, December. [find](https://scholar.google.com/scholar?q=Mike+Thelwall%2C+Kevan+Buckley%2C+Georgios+Paltoglou%2C+Di+Cai%2C+and+Arvid+Kappas%2E+2010%2E+Sentiment+in+Short+Strength+Detection+Informal+Text%2E+Journal+of+the+American+Society+for+Information+Science+and+Techn)
18. Andrea Vanzo, Danilo Croce, and Roberto Basili. 2014. A context-based model for Sentiment Analysis in Twitter. In Proceedings of COLING 2014, the 25th International Conference on Computational Linguistics: Technical Papers, pages 2345–2354, Dublin, Ireland, August. [find](https://scholar.google.com/scholar?q=Andrea+Vanzo%2C+Danilo+Croce%2C+and+Roberto+Basili%2E+2014%2E+A+context%2Dbased+model+for+Sentiment+Analysis+in+Twitter%2E+In+Proceedings+of+COLING+2014%2C+the+25th+International+Conference+on+Computational+Linguis)
19. Theresa Wilson, Janyce Wiebe, and Paul Hoffmann. 2005. [Recognizing Contextual Polarity in Phrase-level Sentiment Analysis](https://md.research.elamur.ai/free/en/papers/wilson-2005-recognizing-contextual-polarity-phrase-level/). In Proceedings of the Conference on Human Language Technology and Empirical Methods in Natural Language Processing, HLT ’05, pages 347–354.

*Updated 2 November 2026*