--- title: "Event Extraction from Unstructured Amharic Text — Tadesse 2020" description: "The authors describe it as the first such work for Amharic. 0.979 recall. 20 references, cited by 2 papers here." authors: ["Ephrem Tadesse", "Rosa Tsegaye", "Kuulaa Qaqqabaa"] year: 2020 venue: "Proceedings of the Twelfth Language Resources and Evaluation Conference" source: "https://aclanthology.org/2020.lrec-1.258/" licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)" licence_url: "https://creativecommons.org/licenses/by-nc/4.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/tadesse-2020-event-extraction-unstructured-amharic-text/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/tadesse-2020-event-extraction-unstructured-amharic-text/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2020](https://md.research.elamur.ai/free/en/papers/2020/page/6/) / Event Extraction from Unstructured Amharic Text # Event Extraction from Unstructured Amharic Text **Authors:** Ephrem Tadesse, Rosa Tsegaye and Kuulaa Qaqqabaa · **Year:** 2020 · **Venue:** Proceedings of the Twelfth Language Resources and Evaluation Conference · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/) **Links:** [Publisher's page](https://aclanthology.org/2020.lrec-1.258/) · [PDF](https://aclanthology.org/2020.lrec-1.258.pdf) · [OpenAlex](https://openalex.org/W3032316664) **Citations:** 2 citations (OpenAlex, 2026-10-14) · 2 in our library, of the 140,884 papers whose bibliographies we read ## Summary The paper addresses event extraction for Amharic, a morphologically rich Semitic language with few NLP resources. The authors describe it as the first such work for Amharic. They assemble a corpus in two ways. First, they scrape about 660K words of news text from several Ethiopian sites and BBC Amharic. Second, they reuse an existing manually POS-tagged news corpus. After heavy preprocessing, a supervised classifier decides whether a candidate expresses an event. The preprocessing includes cleanup, splitting on Amharic punctuation, merging of homophone characters, stemming, stop-word removal and POS tagging. Handcrafted rules then pull out the event and its arguments, with special attention to time expressions. Triggers can be verbs, nouns or even adjectives. Cases the classifier cannot separate cleanly are handed to syntax-based heuristics. Naïve Bayes was the best of three classifiers tested. The hybrid pipeline beat a rule-only baseline, with an F-measure of about 0.97 against about 0.96. The work matters as a template for building event extraction in a low-resource language from scratch. ## Key points - Data comes from two sources: - 659,848 words scraped with Beautiful Soup from Zehabesha, Satenaw, Ezega and BBC Amharic, covering economy, politics, technology and sport. - The Walta Information Center POS-annotated corpus (Demeke and Getachew, 2006): 210,000 words from 1065 news documents. - The preprocessing chain covers cleaning, sentence splitting on Amharic sentence marks, tokenization, stemming, normalization of characters that differ in form but share sound and meaning, stop-word removal and POS tagging. - Three classifiers were compared for event detection: naïve Bayes, SVM and decision tree. Naïve Bayes performed best. - In the reported table, the hybrid system scored precision 0.979, recall 0.962 and F 0.971, against 0.976, 0.952 and 0.959 for the standalone rule-based method. False positives and false negatives were counted by manually scanning the output. - Temporal arguments are hard because Amharic dates may be spelled out in words, written in Arabic numerals or written in Ge'ez numerals, so they need dedicated normalization. ## Interesting ideas - The classifier's confident positive decisions are accepted as they are, while instances that get equal weight for both classes are treated as ambiguous and resolved by syntactic heuristics; this spends rule effort only where the statistical model is unsure. - Event triggers are not restricted to verbs: nouns and even adjectives such as "wedding" or "turmoil" can denote events, and such nominal events may serve as arguments of other events. - Time arguments are normalized across several surface writings (spelled-out words and two numeral systems) before extraction, so that later chronological ordering of events becomes possible. - Orthographic variants that are equivalent in sound and meaning are merged before any lexical processing, which reduces sparsity in a small corpus. - With no labeled event data available, the authors start from a rule-based extractor and combine it with learning, rather than waiting for annotation. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (frames): A low-margin event or role decision could be routed to an explicit, explainable syntactic rule instead of forcing a guess, with the routing itself logged. - **I2** (frames): Frame triggering could cover event nominals, letting a noun open a frame whose roles are filled from its dependents and which can itself fill a role in a verb's frame. - **I3** (world): Time expressions in words or digits could be canonicalized into one value type before events are placed on the world's timeline. - **I4** (tokens): An explicit canonicalization table could be applied ahead of lemma and morphology lookup, covering spelling variants, Unicode look-alikes and British/American forms. - **I5** (learn): Hand rules could serve as a bootstrap witness that produces silver event labels, which a learned component then refines, with the rule-only system kept as a baseline control. *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: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/) Topics: [🔎 Information extraction](https://md.research.elamur.ai/free/en/topics/information-extraction/) · [🐪 Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) ## Cite Cite as: Ephrem Tadesse, Rosa Tsegaye and Kuulaa Qaqqabaa. 2020. Event Extraction from Unstructured Amharic Text. Proceedings of the Twelfth Language Resources and Evaluation Conference. The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{tadesse2020event, title = {{Event Extraction from Unstructured Amharic Text}}, author = {Ephrem Tadesse, Rosa Tsegaye and Kuulaa Qaqqabaa}, booktitle = {Proceedings of the Twelfth Language Resources and Evaluation Conference}, year = {2020}, url = {https://aclanthology.org/2020.lrec-1.258/} } ``` - [← previous Efficient Arabic emotion recognition using deep neural networks](https://md.research.elamur.ai/free/en/papers/ali-2020-efficient-arabic-emotion-recognition-using/) - [next → Is this sentence valid? An Arabic Dataset for Commonsense Validation](https://md.research.elamur.ai/free/en/papers/tawalbeh-2020-this-sentence-valid-arabic-dataset/) - [↑ back to papers of 2020](https://md.research.elamur.ai/free/en/papers/2020/page/6/) ## References 20 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Adafre, S. F. (2005). Part of speech tagging for amharic using conditional random fields. In Proceedings of the ACL workshop on computational approaches to semitic languages. [find](https://scholar.google.com/scholar?q=Adafre%2C+S%2E+F%2E+%282005%29%2E+Part+of+speech+tagging+for+amharic+using+conditional+random+fields%2E+In+Proceedings+of+the+ACL+workshop+on+computational+approaches+to+semitic+languages%2E) 2. Al-Smadi, M. and Qawasmeh, O. (2016). Knowledge-based approach for event extraction from arabic tweets. International Journal of Advanced Computer Science and Applications, 7(6). [find](https://scholar.google.com/scholar?q=Al%2DSmadi%2C+M%2E+and+Qawasmeh%2C+O%2E+%282016%29%2E+Knowledge%2Dbased+approach+for+event+extraction+from+arabic+tweets%2E+International+Journal+of+Advanced+Computer+Science+and+Applications%2C+7%286%29%2E) 3. Arnulphy, B., Claveau, V., Tannier, X., and Vilnat, A. (2015). Supervised Machine Learning Techniques to Detect TimeML Events in French and English. In Chris Beimann, et al., editors, 20thInternational Conference on Applications of Natural Language to Information Systems, NLDB 2015, volume 9103 of Proceedings of the NLDB conference, Passau, Germany, June. Springer. [find](https://scholar.google.com/scholar?q=Arnulphy%2C+B%2E%2C+Claveau%2C+V%2E%2C+Tannier%2C+X%2E%2C+and+Vilnat%2C+A%2E+%282015%29%2E+Supervised+Machine+Learning+Techniques+to+Detect+TimeML+Events+in+French+and+English%2E+In+Chris+Beimann%2C+et+al%2E%2C+editors%2C+20thInternationa) 4. Baradaran, R. and Mineai-Bidgoli, B. (2015). Event extraction from classical arabic texts. International Arab Journal of Information Technology", 12(5). [find](https://scholar.google.com/scholar?q=Baradaran%2C+R%2E+and+Mineai%2DBidgoli%2C+B%2E+%282015%29%2E+Event+extraction+from+classical+arabic+texts%2E+International+Arab+Journal+of+Information+Technology%22%2C+12%285%29%2E) 5. Chang, C.-C. and Lin, C.-J. (2011). Libsvm: A library for support vector machines. ACM Trans. Intell. Syst. Technol., 2(3):27:1–27:27, May. [find](https://scholar.google.com/scholar?q=Chang%2C+C%2E%2DC%2E+and+Lin%2C+C%2E%2DJ%2E+%282011%29%2E+Libsvm%3A+A+library+for+support+vector+machines%2E+ACM+Trans%2E+Intell%2E+Syst%2E+Technol%2E%2C+2%283%29%3A27%3A1%E2%80%9327%3A27%2C+May%2E) 6. Demeke, G. and Getachew, M. (2006). Manual annotation of amharic news items with part-of-speech tags and its challenges. 01. [find](https://scholar.google.com/scholar?q=Demeke%2C+G%2E+and+Getachew%2C+M%2E+%282006%29%2E+Manual+annotation+of+amharic+news+items+with+part%2Dof%2Dspeech+tags+and+its+challenges%2E+01%2E) 7. Frederik Hogenboom, F. F. and Kaymak, U. (2016). A survey of event extraction methods from text for decision support systems. Elsevier. [find](https://scholar.google.com/scholar?q=Frederik+Hogenboom%2C+F%2E+F%2E+and+Kaymak%2C+U%2E+%282016%29%2E+A+survey+of+event+extraction+methods+from+text+for+decision+support+systems%2E+Elsevier%2E) 8. Gasser, M. (2011). Hornmorpho: a system for morphological processing of amharic, oromo, and tigrinya. In Conference on Human Language Technology for Development. [find](https://scholar.google.com/scholar?q=Gasser%2C+M%2E+%282011%29%2E+Hornmorpho%3A+a+system+for+morphological+processing+of+amharic%2C+oromo%2C+and+tigrinya%2E+In+Conference+on+Human+Language+Technology+for+Development%2E) 9. Ibrahim, A. and Assabie, Y. (2014). Amharic sentence parsing using base phrase chunking. In COLING 2014. lasker, L., Argaw, A. A., and Gamback, B. (2007). Applying machine learning to amharic text classification. In Proceedings of the 5th World Congress of African Linguistics. [find](https://scholar.google.com/scholar?q=Ibrahim%2C+A%2E+and+Assabie%2C+Y%2E+%282014%29%2E+Amharic+sentence+parsing+using+base+phrase+chunking%2E+In+COLING+2014%2E+lasker%2C+L%2E%2C+Argaw%2C+A%2E+A%2E%2C+and+Gamback%2C+B%2E+%282007%29%2E+Applying+machine+learning+to+amharic+text+cla) 10. Mulugeta, W. and Gasser, M. (2012). Learning morphological rules for amharic verbs using inductive logic programming. [find](https://scholar.google.com/scholar?q=Mulugeta%2C+W%2E+and+Gasser%2C+M%2E+%282012%29%2E+Learning+morphological+rules+for+amharic+verbs+using+inductive+logic+programming%2E) 11. Pranckevicius, T. and Marcinkevicius, V. (2017). Comparison of naïve bayes, random forest, decision tree, support vector machines, and logistic regression classifiers for text reviews classification. In Baltic J. Modern Computing. [find](https://scholar.google.com/scholar?q=Pranckevicius%2C+T%2E+and+Marcinkevicius%2C+V%2E+%282017%29%2E+Comparison+of+na%C3%AFve+bayes%2C+random+forest%2C+decision+tree%2C+support+vector+machines%2C+and+logistic+regression+classifiers+for+text+reviews+classification%2E+) 12. Ramesh, D. and Kumar, S. S. (2016). Event extraction from natural language text. International Journal of Engineering Sciences and Research Technology (IJESRT), 5(7). [find](https://scholar.google.com/scholar?q=Ramesh%2C+D%2E+and+Kumar%2C+S%2E+S%2E+%282016%29%2E+Event+extraction+from+natural+language+text%2E+International+Journal+of+Engineering+Sciences+and+Research+Technology+%28IJESRT%29%2C+5%287%29%2E) 13. Ramrakhiyani, N. and Majumder. (2015). Approaches to temporal expression recognition in hindi. ACM Transactions on Asian and Low-Resource Language Information Processing (TALLIP), 14(1). [find](https://scholar.google.com/scholar?q=Ramrakhiyani%2C+N%2E+and+Majumder%2E+%282015%29%2E+Approaches+to+temporal+expression+recognition+in+hindi%2E+ACM+Transactions+on+Asian+and+Low%2DResource+Language+Information+Processing+%28TALLIP%29%2C+14%281%29%2E) 14. Schmid, H. (1994). Probabilistic part-of-speech tagging using decision trees. In International Conference on New Methods in Language Proceeding. [find](https://scholar.google.com/scholar?q=Schmid%2C+H%2E+%281994%29%2E+Probabilistic+part%2Dof%2Dspeech+tagging+using+decision+trees%2E+In+International+Conference+on+New+Methods+in+Language+Proceeding%2E) 15. Sikdar, U. and Gambäck, B., (2018). Named Entity Recognition for Amharic Using Stack-Based Deep Learning: 18th International Conference, CICLing 2017, Budapest, Hungary, April 17–23, 2017, Revised Selected Papers, Part I, pages 276–287. 01. [find](https://scholar.google.com/scholar?q=Sikdar%2C+U%2E+and+Gamb%C3%A4ck%2C+B%2E%2C+%282018%29%2E+Named+Entity+Recognition+for+Amharic+Using+Stack%2DBased+Deep+Learning%3A+18th+International+Conference%2C+CICLing+2017%2C+Budapest%2C+Hungary%2C+April+17%E2%80%9323%2C+2017%2C+Revised+Sel) 16. Smadi, M. and Qawasmeh, O. (2018). A supervised machine learning approach for events extraction out of arabic tweets. In Fifth International Conference on Social Networks Analysis, Management and Security, SNAMS 2018, Valencia, Spain, October 15-18, 2018, pages 114– 119. [find](https://scholar.google.com/scholar?q=Smadi%2C+M%2E+and+Qawasmeh%2C+O%2E+%282018%29%2E+A+supervised+machine+learning+approach+for+events+extraction+out+of+arabic+tweets%2E+In+Fifth+International+Conference+on+Social+Networks+Analysis%2C+Management+and+Secu) 17. Sohail, O. and Elahi, I. (2018). Text classification in an under-resourced language via lexical normalization and feature pooling. In Twenty-Second Pacific Asia Conference on Information Systems. [find](https://scholar.google.com/scholar?q=Sohail%2C+O%2E+and+Elahi%2C+I%2E+%282018%29%2E+Text+classification+in+an+under%2Dresourced+language+via+lexical+normalization+and+feature+pooling%2E+In+Twenty%2DSecond+Pacific+Asia+Conference+on+Information+Systems%2E) 18. Tourille, J., Ferret, O., Tannier, X., and Neveol, A. (2017). Temporal information extraction from clinical text. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics, volume 2 of EACL, page 739–745. [find](https://scholar.google.com/scholar?q=Tourille%2C+J%2E%2C+Ferret%2C+O%2E%2C+Tannier%2C+X%2E%2C+and+Neveol%2C+A%2E+%282017%29%2E+Temporal+information+extraction+from+clinical+text%2E+In+Proceedings+of+the+15th+Conference+of+the+European+Chapter+of+the+Association+for+C) 19. Tsedalu, G. (2010). Information extraction model from amharic news texts. Master’s thesis, Addis Ababa University. [find](https://scholar.google.com/scholar?q=Tsedalu%2C+G%2E+%282010%29%2E+Information+extraction+model+from+amharic+news+texts%2E+Master%E2%80%99s+thesis%2C+Addis+Ababa+University%2E) 20. Yunita Sari, M. F. H. and Zamin, N. (2010). Rule based pattern extractor and named entity recognition: A hybrid approach. IEEE. [find](https://scholar.google.com/scholar?q=Yunita+Sari%2C+M%2E+F%2E+H%2E+and+Zamin%2C+N%2E+%282010%29%2E+Rule+based+pattern+extractor+and+named+entity+recognition%3A+A+hybrid+approach%2E+IEEE%2E)