Elamur Research⌕
☰
home / papers / 2002 / QARAB: A: Question Answering System to Support the Arabic Language

QARAB: A: Question Answering System to Support the Arabic Language

Paper · 2002 · Bassam Hammo, Hani Abu-Salem, Steven Lytinen et al.

Summary

QARAB addresses open-domain question answering for Arabic, where NLP lagged far behind English. The authors name several reasons: rich inflection and derivation, unwritten vowels (diacritics), no capital letters to signal proper names, and few corpora and lexicons. The system answers from a collection of articles from the Qatari newspaper Al-Raya.

A retrieval engine built from scratch on Salton's vector space model is stored as relational database tables. It treats the question as a query, widens it through shared Arabic roots, and returns ranked documents split into paragraphs. The question and the candidate paragraphs are then processed the same way. Stop-words are dropped, words are stemmed and tagged with an existing tagger and proper-name finder, and the question word fixes the expected answer type (for example, "who" means a person). A passage is kept when it contains the question's nouns in their original order plus cue words for the expected entity type.

No quantitative evaluation is reported, only a promising informal trial. Still, the paper is an early blueprint for Arabic QA that adds lightweight linguistic matching on top of IR.

Key points

  • Knowledge source: Al-Raya (Qatar) newspaper text. The system assumes each answer lies within a single document and is a short passage.
  • The IR index is a set of relational tables: ROOT_TABLE, STEM_TABLE (with document frequency and IDF), POSTING_TABLE (document, paragraph, position, length), DOCUMENT_TABLE, PARAGRAPH_TABLE and QUERY_TABLE (original and root-expanded words with weights).
  • Part-of-speech and proper-name tagging rely on the Type-Finder and Proper Name-Finder by Abuleil [1999]. Verbs are reported as easiest to spot thanks to regular morphological patterns.
  • Hand-built Arabic keyword lists help detect person names, organizations, locations, numbers, money and dates.
  • The pipeline has three stages: analyse the question, retrieve candidate paragraphs, then analyse those paragraphs like the question and return sentences likely to hold the answer. The planned evaluation has Arabic users supply questions and judge whether answers satisfy them.

Interesting ideas

  • I1. The interrogative word sets an expected answer category, and small cue-word lists then locate candidates of that category in a passage, a cheap and transparent typed-answer filter.
  • I2. Requiring the question's nouns to appear in the candidate passage in the same order is a minimal structural check that sits above pure keyword overlap.
  • I3. Query expansion through the morphological root links a question word to every stem in the same derivational family, not just its exact form.
  • I4. Storing the inverted index with paragraph IDs, token positions and lengths in plain relational tables makes every retrieval hit traceable to an exact text span.
  • I5. The authors plan to collect real users' question forms and let those same users judge answer satisfaction, rather than relying only on gold strings.

For Mova

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

  • I1 (answer): Readers answering fable or school-test questions could map wh-words to an expected entity or value type and reject candidate answers whose world entity has the wrong type, with the rejection reason logged.
  • I2 (answer): It could serve as a weak baseline or negative control for the tree- and frame-based answerer, showing which questions really need structure rather than word order.
  • I3 (tokens): A derivational-family layer over English lemmas (e.g., decide/decision) would let the matcher connect a nominal question to a verbal sentence while recording which derivation was used.
  • I4 (world): Keeping span provenance on each entity and event in the world of the text would let every answer point back to the paragraph and tokens that support it.
  • I5 (learn): Evaluation sets for the QA readers could include human-authored question variants and acceptability judgements, catching answers that are technically correct but unhelpful.

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

🎓DePaul UniversityChicago · United States2 works · 148 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksInformation Systems · 1,437 works

🧭 Topics 2Research subjects and languages the work is about.Semitic languages · 2,908 worksQuestion answering · 441 works

Cite

Cite as Bassam Hammo, Hani Abu-Salem, Steven Lytinen and Martha Evens. 2002. QARAB: A: Question Answering System to Support the Arabic Language. Proceedings of the ACL-02 Workshop on Computational Approaches to Semitic Languages.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/1118637.1118644

@inproceedings{hammo2002qarab,
  title = {{QARAB: A: Question Answering System to Support the Arabic Language}},
  author = {Bassam Hammo, Hani Abu-Salem, Steven Lytinen and Martha Evens},
  booktitle = {Proceedings of the ACL-02 Workshop on Computational Approaches to Semitic Languages},
  year = {2002},
  doi = {10.3115/1118637.1118644},
  url = {https://doi.org/10.3115/1118637.1118644}
}

References

37 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Abuleil, S., and Evens, M., 1998. “Discovering Lexical Information by Tagging Arabic Newspaper Text”, Workshop on Semantic Language Processing. COLING-ACL ’98, University of Montreal, Montreal, PQ, Canada, Aug. 16 1998, pp. 1-7.
  2. Al-Daimi, K., and Abdel-Amir, M. 1994. “The Syntactic Analysis of Arabic by Machine”. Computers and Humanities, Vol. 28, No. 1, pp. 29-37. ↗
  3. Allan, J., Callan, J., Feng, F-F., and Malin D. 1999. “INQUERY and TREC-8”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 637-645. find
  4. Ask Jeeves. 1996. www.ask.com Site last visited in March 2001. find
  5. Breck, E., Burger, J., Ferro, L., House, D., Light, M., and Mani, I. 1999. “A Sys Called Qanda”. Proceedings of the 8th Text REtrieval Conference, NIST Special Publications, pp. 499-507. find
  6. Budzik, J. and Hammond, K. 1999. “Q&A: A System for the Capture, Organization and Reuse of Expertise”. Proceedings of the Sixty-second Annual Meeting of the American Society for Information Science. Information Today, Inc., Medford, NJ. Available on the Web at http://dent.infolab.nwu.edu/infolab/downloads/papers/paper10061.pdf. Site last visited in August 2001. ↗
  7. Burke, R., Hammond, K., Kulyukin, V., Lytinen, S., Tomuro, N., and Schoenberg, S. 1997. “Question Answering from Frequently-Asked Question Files: Experiences with the FAQ Finder System”. AI Magazine, Vol. 18, No.2, pp. 57-66. ↗
  8. Cardie, C., Ng, V., Pierce, D., and Buckley, C. 2000. “Examining the Role of Statistical and Linguistic Knowledge Sources in a General-Knowledge Question-Answering System”. Proceedings of the Sixth Applied Natural Language Processing Conference, pp. 180-187. find
  9. Cormack, G., Clarke, C., and Kisman, D. 1999. “Fast Automatic Passage Ranking (MultiText Experiments for TREC-8)”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 735-743. find
  10. Ferret, O., Grau, B., Illouz, G., Jacquemin, C., and Masson, N. 1999. “QALC - the Question-Answering Program of the Language and Cognition Group at LIMSI-CNRS”. Proceedings of the 8th Text REtrieval Conference, NIST Special Publications, pp. 465-475. find
  11. Harabagiu, S., Pasca, M., and Maiorano, S. 2000. “Experiments with Open-Domain Textual Question Answering”. Proceedings of 18 th International Conference on Computational Linguistics (COLING-2000), Saarbrucken, Germany, pp. 292-298 find
  12. Hull, D. 1999. “Xerox TREC-8 Question Answering Track Report”. Proceedings of the 8 th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 743-751. find
  13. Humphreys, K., Gaizauskas, R., Hepple, M., and Sanderson, M. 1999. “University of Sheffield TREC-8 Q & A System”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 707-717. find
  14. Jacobs, P., and Rau, L. 1990. “SCISOR: Extracting Information from On-line News”. Communications of the ACM, Vol. 33, No.11, pp. 88-97. ↗
  15. Katz, B. 1997. “From Sentence Processing to Information Access on the World Wide Web”. Proceedings of the American Association for Artificial Intelligence Conference, Spring Symposium, NLP for WWW, pp. 77-86. find
  16. Khoja, S. 1999. “Stemming Arabic Text”. Available on the Web at: http://www.comp.lancs.ac.uk/computing/users/khoja/stemmer.ps. Site last visited in March 2001. ↗
  17. Kupiec, J. 1993. “MURAX: A Robust Linguistic Approach for Question Answering Using an On-line Encyclopedia”. Proceedings of the 16th Annual Int. ACM SIGIR Conference, pp. 181-190. find
  18. Lehnert, W. 1978. The Process of Question Answering. Lawrence Erlbaum Associates, Hillsdale, NJ. find
  19. Lin, C-J, and Chen, H-H. 1999. “Description of Preliminary Results to TREC-8 QA Task”. Proceedings of the 8th Text REtrieval Conference(TERC-8), NIST Special Publications 500-246, pp. 507-513. find
  20. Litkowski, K. 1999. “Question-Answering Using Semantic Relation Triples”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-248, pp. 349-357 find
  21. Lundquist, C., Grossman, D., and Frieder, O. 1999. "Improving Relevance Feedback in the Vector Space Model". Proceedings of 6 th ACM Annual Conference on Information and Knowledge Management (CIKM), pp. 16-23. find
  22. Moldovan, D., Harabagiu, S., Pasca, M., Mihalcea, R., Girju, R., Goodrum, R., and Rus, V. 2000. “The Structure and Performance of an Open-Domain Question-Answering System”. Proceedings of the 38 th Annual Meeting of the Association for Computational Linguistics, pp. 563-570. find
  23. Oard, D., Wang, J., Lin, D., and Soboroff, I. 1999. “TREC-8 Experiments at Maryland: CLIR, QA and Routing”. Proceedings of the 8th Text REtrieval Conference (TERC-8), NIST Special Publications 500-246, pp. 623-637. ↗
  24. Ogden, B., Cowie, J., Ludovik, E. Molina-Salgado, H., Nirenburg, S., Sharples, N., and Sheremtyeva, S. 1999. “CRL's TREC-8 Systems Cross-Lingual IR, and Q&A”. Proceedings of the 8th Text REtrieval Conference (TERC-8), NIST Special Publications 500-246, pp. 513-523. find
  25. Salton, G. 1971. The SMART Retrieval System Experiments in Automatic Document Processing. Prentice Hall Inc., Englewood Cliffs, NJ. find
  26. Schank, R., and Abelson, R. 1977. Scripts, Plans, Goals, and Understanding. Lawrence Erlbaum Associates, Hillsdale, NJ. find
  27. Shin, D-H, Kim, Y-H, Kim, S., Eom, J-H, Shin, H-J, and Zhang B-T. 1999. “SCAI TREC-8 Experiments”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 583-591. find
  28. Singhal, A., Abney, S., Bacchiani, M., Collins, M., Hindle, D., and Pereira, F. 1999. “AT&T at TREC-8”. Proceedings of the 8th Text REtrieval Conference, NIST Special Publications, pp. 317-331. find
  29. Srihari, R., and Li, W. 1999. “Information Extraction Supported Question Answering”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 185-197. ↗
  30. Takaki, T. 1999. “NTT DATA: Overview of System Approach at TREC-8 ad-hoc and Question Answering”. Proceedings of the 8th Text REtrieval Conference (TREC-8), NIST Special Publications 500-246, pp. 523-531. find
  31. TREC-8. 1999. NIST Special Publication 500-246: The Eighth Text REtrieval Conference. Available on the Web at: http://trec.nist.gov/pubs/trec8/t8_proceedings. html. Site last visited in August 2001. ↗
  32. TREC-9. 2000. NIST Special Publication: The Ninth Text REtrieval Conference. Available on the Web at: http://trec.nist.gov/pubs/trec9/t9_proceedings. html. Site last visited in August 2001. ↗
  33. Vicedo, J., and Ferrández, A. 2000. “Importance of Pronominal Anaphora Resolution in Question- Answering System”. Proceedings of the 38 th Annual Meeting of the Association for Computational Linguistics, pp. 555-562. find
  34. Voorhees, E., and Tice, D. 1999. "The TREC-8 Question Answering Track Evaluation". Proceedings of the 8 th Text REtrieval Conference (TREC-8), NIST Special Publication 500-246, pp. 83-106.
  35. Voorhees, E., and Tice, D. 2000. “Building a Question Answering Test Collection”. Proceedings of the 23 rd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Athens, Greece, pp. 200-207. find
  36. Winograd, T. 1972. Understanding Natural Language. Academic Press, New York, NY. ↗
  37. Woods, W., Kaplan, R., and Webber, B. 1972. “The Lunar Sciences Natural Language Information System: Final Report”. Bolt Beranek and Newman Inc. (BBN), Report No. 2378. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/hammo-2002-qarab-question-answering-system-support/