Elamur Research⌕
☰
home / papers / 2019 / OlloBot - Towards A Text-Based Arabic Health Conversational Agent: Evaluation and Results

OlloBot - Towards A Text-Based Arabic Health Conversational Agent: Evaluation and Results

Paper · 2019 · Ahmed Fadhil and Ahmed AbuRa'ed

Summary

OlloBot is a text chatbot for Arabic speakers who cannot use English-only health tools. It supports patients and the doctors who care for them. The authors justify it with the heavy burden of noncommunicable diseases in Arabic-speaking countries and the shortage of health assistants in Arabic. The bot covers healthy eating, physical activity, mental wellbeing and coping skills, and it can log daily meals. Its dialogue states, intents and entities are built on IBM Watson Conversation. Users talk to it through Telegram, and a Node.js layer handles dialogue logic the flow does not match. The conversation design came from a dietitian working with a clinic, and that guidance was translated into Arabic. Interaction data are stored as reports for caregivers. Forty-three Arabic-speaking participants tested the bot and rated it with the USE questionnaire. All four dimensions scored significantly above the neutral midpoint, with ease of learning rated highest. The work matters as an early user evaluation of a language-specific health agent for an under-served language community.

Key points

  • The system is a Telegram front end, IBM Watson Conversation for NLU and dialogue flow, and a Node.js wrapper for unmatched turns. Logged data are saved as reports caregivers can consult.
  • The intents are greetings, coping skills, diet/nutrition, daily meal logging, mood, jokes, physical activity, conditions, farewells and an "anything else" fallback.
  • The entities include time words, quantities/units (kg, g, teaspoon, bread loaf), numbers, a list of common Arabic foods and sports. Each entity has synonym lists.
  • In the USE questionnaire (n=43, one-sample t-tests against 3, df=42, all p<.01), the means were:
    • Usefulness 3.381
    • Ease of Use 3.626
    • Ease of Learning 3.942
    • Satisfaction 3.505
  • Results did not differ by gender or by use of tracking devices. Users unfamiliar with chatbots gave lower learnability scores (t(41) = −2.46). The chatbot-vs-app preference averaged 3.86 on a 1–5 scale, leaning toward the chatbot.

Interesting ideas

  • I1. Users can log a meal as typed items or as a string of emoji, and each emoji is resolved to a food item. This treats pictographs as real lexical content rather than noise.
  • I2. Quantity/unit mentions are kept as an entity separate from bare countable numbers, which keeps "two" apart from "teaspoon".
  • I3. The bot has an explicit catch-all intent for input it cannot handle, instead of forcing the closest known intent.
  • I4. Users can switch topics while the bot is still waiting for slot values, and the dialogue state is kept so they can come back to an unfinished task later.
  • I5. Ratings are tested against the neutral scale midpoint, then split by user groups such as prior chatbot familiarity, which exposes effects that an overall average would hide.

For Mova

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

  • I1 (tokens): the tokenizer could give emoji their own token class with a lemma and semantic class, so later stages can read them as nouns instead of dropping them.
  • I2 (world): measure handling could keep separate slots for count, unit and substance, so phrases like "two teaspoons of sugar" become one structured holding.
  • I3 (answer): readers could return an explicit "not covered" outcome when no frame matches, which keeps failures visible and explainable.
  • I4 (frames): pending speech acts such as open questions could be stored as unfinished frames, so an interruption is linked back when the topic resumes.
  • I5 (learn): evaluations could always report scores against a neutral baseline and broken down by input subgroup, acting as a cheap negative control for accuracy gains that are not real.

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 TrentoTrento · Italy85 works · 15,524 citations🎓Universitat Pompeu FabraBarcelona · Spain47 works · 11,723 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Medicine · 241 worksHealth Informatics · 184 worksHealth Professions · 69 worksHealth Information Management · 50 worksGeneral Health Professions · 14 works

🧭 Topics 2Research subjects and languages the work is about.Semitic languages · 2,908 worksDialogue systems · 624 works

Cite

Cite as Ahmed Fadhil and Ahmed AbuRa'ed. 2019. OlloBot - Towards A Text-Based Arabic Health Conversational Agent: Evaluation and Results. Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2019).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.26615/978-954-452-056-4_034

@inproceedings{fadhil2019ollobot,
  title = {{OlloBot - Towards A Text-Based Arabic Health Conversational Agent: Evaluation and Results}},
  author = {Ahmed Fadhil and Ahmed AbuRa'ed},
  booktitle = {Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2019)},
  year = {2019},
  doi = {10.26615/978-954-452-056-4\_034},
  url = {https://doi.org/10.26615/978-954-452-056-4_034}
}

References

26 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Gillian E Mead, Wendy Morley, Paul Campbell, Carolyn A Greig, Marion McMurdo, Debbie A Lawlor, et al. 2009. Exercise for depression. Cochrane Database Syst Rev 3. ↗
  2. Bayan Aref Abu Shawar. 2005. A corpus based approach to generalise a chatbot system. Ph.D. thesis, University of Leeds. find
  3. Susan Michie, Charles Abraham, Craig Whittington, John McAteer, and Sunjai Gupta. 2009. Effective techniques in healthy eating and physical activity interventions: a meta-regression. Health Psychology 28(6):690. ↗
  4. Abdulaziz M Alayba, Vasile Palade, Matthew England, and Rahat Iqbal. 2017. Arabic language sentiment analysis on health services. arXiv preprint arXiv:1702.03197 . ↗
  5. Dana Abu Ali and Nizar Habash. 2016. Botta: An arabic dialect chatbot. In COLING (Demos). pages 208–212.
  6. World Health Organization et al. 2006. Global strategy on diet, physical activity and health: a framework to monitor and evaluate implementation . find
  7. Michal Artal, Carl Sherman, and Nicholas A DiNubile. 1998. Exercise against depression. The Physician and sportsmedicine 26(10):55–70. ↗
  8. World Health Organization et al. 2010. WORLD HEALH REPORT (The): Health Systems Financing: the path to universal Coverage (Arabic). World Health Organization. find
  9. A Byrne and DG Byrne. 1993. The effect of exercise on depression, anxiety and other mood states: a review. Journal of psychosomatic research 37(6):565–574. ↗
  10. World Health Organization et al. 2012. World health day 2012: ageing and health: toolkit for event organizers . find
  11. Gary Cooney, Kerry Dwan, and Gillian Mead. 2014. Exercise for depression. Jama 311(23):2432–2433. ↗
  12. Ruth A Parslow and Anthony F Jorm. 2000. Who uses mental health services in australia? an analysis of data from the national survey of mental health and wellbeing. Australian and New Zealand Journal of Psychiatry 34(6):997–1008. ↗
  13. Kathleen Kara Fitzpatrick, Alison Darcy, and Molly Vierhile. 2017. Delivering cognitive behavior therapy to young adults with symptoms of depression and anxiety using a fully automated conversational agent (woebot): A randomized controlled trial. JMIR Mental Health 4(2):e19. ↗
  14. Barbara Olasov Rothbaum, Elizabeth A Meadows, Patricia Resick, and David W Foy. 2000. Cognitive-behavioral therapy. . find
  15. Bettina Graf, Maike Krüger, Felix Müller, Alexander Ruhland, and Andrea Zech. 2015. Nombot: simplify food tracking. In Proceedings of the 14th International Conference on Mobile and Ubiquitous Multimedia. ACM, pages 360–363. find
  16. B Abu Shawar and Eric Atwell. 2009. Arabic question-answering via instance based learning from an faq corpus. In Proceedings of the CL 2009 International Conference on Corpus Linguistics. UCREL. volume 386. ↗
  17. Eat Smart Play Hard, Healthy Kansas Schools, In-Class Physical Activity Breaks, and Blast Off Game. 2012. Physical activity. Fitness and health: Internatio . find
  18. Martin Sundermeyer, Tamer Alkhouli, Joern Wuebker, and Hermann Ney. 2014. Translation modeling with bidirectional recurrent neural networks. In EMNLP. pages 14–25. find
  19. Saliza Ismail and A Ahmad. 2004. Recurrent neural network with back propagation through time algorithm for arabic recognition. Proceedings of the 18th ESM Magdeburg, Germany pages 13–16. find
  20. Theodore B Van Itallie. 1985. Health implications of overweight and obesity in the united states. Annals of internal medicine 103(6 Part 2):983–988. ↗
  21. Ronald C Kessler and T Bedirhan Üstün. 2004. The world mental health (wmh) survey initiative version of the world health organization (who) composite international diagnostic interview (cidi). International journal of methods in psychiatric research 13(2):93– 121. ↗
  22. World Health Organization (WHO et al. 2017. Obesity and overweight factsheet from the who. Health . find
  23. Wajdi Zaghouani, Nizar Habash, Houda Bouamor, Alla Rozovskaya, Behrang Mohit, Abeer Heider, and Kemal Oflazer. 2015. Correction annotation for non-native arabic texts: Guidelines and corpus. In LAW@ NAACL-HLT. pages 129–139.
  24. Hian Chye Koh, Gerald Tan, et al. 2011. Data mining applications in healthcare. Journal of healthcare information management 19(2):65. find
  25. Arnold M Lund. 2001. Measuring usability with the use questionnaire12. Usability interface 8(2):3–6. find
  26. Traci Mann, A Janet Tomiyama, Erika Westling, Ann-Marie Lew, Barbra Samuels, and Jason Chatman. 2007. Medicare’s search for effective obesity treatments: diets are not the answer. American Psychologist 62(3):220. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/fadhil-2019-ollobot-text-based-arabic-health/