Elamur Research⌕
☰
home / papers / 2013 / An NLP-based Reading Tool for Aiding Non-native English Readers

An NLP-based Reading Tool for Aiding Non-native English Readers

Paper · 2013 · Mahmoud Azab, Ahmed Salama, Kemal Oflazer et al.

Summary

The paper presents a browser-based e-book reader that helps non-native English readers, especially students, get past vocabulary, grammar and context problems while reading. Instead of building new models, the authors combine many off-the-shelf English NLP components into one client-server application built around UIMA. Texts from a library, or texts the user uploads, are annotated ahead of time on the server. The browser client is kept thin and only sends queries about what the reader clicked or selected. For each click, the system guesses what the reader most likely wants and shows that as the default answer, with a menu of other options. These include a word's meaning, its grammatical role, the full named entity, a pronoun's antecedent and other mentions, auto-generated questions with answers, and summaries. A small manual check on one document of about 1000 words gives per-component precision and recall. User studies were still being planned. The work shows that existing NLP tools can be assembled into a practical reading aid.

Key points

  • The architecture is UIMA-based. All annotations except summaries are stored in a UIMA file and retrieved through UIMA's query mechanisms. The authors describe the design as language-independent as long as UIMA-compatible annotators exist.
  • Components used:
    • Stanford CoreNLP for dependencies, NER and coreference
    • WordNet for meanings, using the first sense after morphological analysis and POS filtering
    • Heilman and Smith's question generator for in-text QA
    • the MEAD toolkit for paragraph-, chapter- and document-level summaries
  • Questions about an entity are generated from earlier sentences during annotation. They are ranked by length, proximity, and whether the question or its answer involves another named entity.
  • Manual intrinsic evaluation on a document of about 1000 words:
    • NER: P 0.909, R 0.869, F 0.888
    • POS: P 0.986
    • coreference: P 0.679, R 0.63, F 0.653
    • word meaning: P 0.861, R 0.831, F 0.845
    • in-text QA: P 0.62
  • Planned work includes combining several WSD systems with super-sense taggers to beat the most-frequent-sense baseline, a lexical simplification module, and an extrinsic study with a critical reading course.

Interesting ideas

  • I1. Intent-ranked responses. The type of annotation at the click position (content word, function word, entity, pronoun) decides which answer is shown first, and the rest stay one menu step away; this keeps a rich interface from overwhelming the user
  • I2. Friendly dependency labels. Raw dependency labels are translated into plain-language role descriptions before being shown to learners, which makes parser output understandable to non-linguists
  • I3. Expanding a click to the full unit. Selecting one word highlights the whole span it belongs to, such as a named entity or a possibly discontinuous phrasal verb like break through, and gives the meaning of both the whole unit and the single word
  • I4. Precomputed questions per entity, ranked by simple cues. Question–answer pairs about each entity are generated in advance from preceding text and then ranked by length, proximity and the involvement of other entities
  • I5. Mention navigation for non-linear reading. Coreference chains let the reader jump back to an entity's first mention or nearby ones and then return to where they were

For Mova

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

  • I1 (answer): Explanations of a reading could lead with the layer most relevant to the clicked token, such as its frame role for a verb argument or its antecedent for a pronoun, with the full UD tree and world state offered as drill-down.
  • I2 (tree): A mapping from UD relations to readable phrases would make tree-level explanations usable for students and for checking parses.
  • I3 (tokens): Treating discontinuous phrasal verbs and compounds as one lexical unit, while still keeping per-token lemmas, would give later stages the right meaning-bearing item.
  • I4 (learn): Questions generated from earlier sentences about tracked entities could serve as self-play probes of the world model, with cheap ranking cues used to choose which ones to keep.
  • I5 (world): Exposing each entity's mention chain as a navigable, explainable trace would make coreference decisions in the text's world easy to inspect and debug.

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, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Mahmoud Azab, Ahmed Salama, Kemal Oflazer, Hideki Shima, Jun Araki and Teruko Mitamura. 2013. An NLP-based Reading Tool for Aiding Non-native English Readers. Proceedings of the International Conference Recent Advances in Natural Language Processing RANLP 2013.

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

@inproceedings{azab2013nlp,
  title = {{An NLP-based Reading Tool for Aiding Non-native English Readers}},
  author = {Mahmoud Azab, Ahmed Salama, Kemal Oflazer, Hideki Shima, Jun Araki and Teruko Mitamura},
  booktitle = {Proceedings of the International Conference Recent Advances in Natural Language Processing RANLP 2013},
  year = {2013},
  url = {https://aclanthology.org/R13-1006/}
}

References

24 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mahmoud Azab, Ahmed Salama, Kemal Oflazer, Hideki Shima, Jun Araki, and Teruko Mitamura. 2013. An English reading tool as a NLP showcase. In Proceedings of IJCNLP – System Demonstration, Nagoya, Japan.
  2. Massimiliano Ciaramita and Yasemin Altun. 2006. Broad-coverage sense disambiguation and information extraction with a supersense sequence tagger. In Proceedings of EMNLP’06, EMNLP ’06, pages 594–602, Stroudsburg, PA, USA.
  3. Marie-Catherine De Marneffe, Bill MacCartney, and Christopher D Manning. 2006. Generating typed dependency parses from phrase structure parses. In Proceedings of LREC, volume 6, pages 449–454.
  4. Margaret M Dermody and Richard B Speaker Jr. 1999. Reciprocal Strategy Training in Prediction, Clarification, Question Generating and Summarization to Improve Reading Comprehension. Reading Improvement, 36(1):16–23. find
  5. Soojeong Eom, Markus Dickinson, and Rebecca Sachs. 2012. Sense-specific lexical information for reading assistance. In Proceedings of the Seventh Workshop on Building Educational Applications Using NLP, pages 316–325, Stroudsburg, PA, USA. find
  6. Rachele De Felice. 2008. Automatic Error Detection in Non-native English. Ph.D. thesis, St Catherines College, University of Oxford. ↗
  7. Christiane Fellbaum. 1998. WordNet: An electronic lexical database. The MIT Press. ↗
  8. David Ferrucci and Adam Lally. 2004. UIMA: an architectural approach to unstructured information processing in the corporate research environment. Natural Language Engineering, 10(3-4):327–348. ↗
  9. Marie-Jose Hamel and Marie-Christine Girard. 2000. FreeText - an advanced hypermedia CALL system featuring NLP tools for a smart treatment of authentic documents and free production exercises. In Proceedings of EuroCALL-2000. find
  10. Michael Heilman and Noah Smith. 2010. Extracting simplified statements for factual question generation. In Proceedings of the 3rd Workshop on Question Generation. find
  11. Michael Heilman, Kevyn Collins-Thompson, Jamie Callan, and Maxine Eskenazi. 2006. Classroom success of an intelligent tutoring system for lexical practice and reading comprehension. In Proceedings of the Ninth International Conference on Spoken Language Processing. find
  12. Heeyoung Lee, Yves Peirsman, Angel Chang, Nathanael Chambers, Mihai Surdeanu, and Dan Jurafsky. 2011. Stanford’s multi-pass sieve coreference resolution system at the conll shared task. In Proceedings CONLL’11, pages 28–34. find
  13. Heeyoung Lee, Angel Chang, Yves Peirsman, Nathanael Chambers, Mihai Surdeanu, and Dan Jurafsky. 2013. Deterministic coreference resolution based on entity-centric, precision-ranked rules. Computational Linguistics, pages 1–54. ↗
  14. Stuart Luppescu and Richard R. Day. 1992. Reading, Dictionaries, and Vocabulary Learning. Language Learning, 43(2):263–279. ↗
  15. Mohamed Maamouri, Wajdi Zaghouani, Violetta Cavalli-Sforza, Dave Graff, and Mike Ciul. 2012. Developing ARET: an NLP-based educational tool set for Arabic reading enhancement. In Proceedings of the Seventh Workshop on Building Educational Applications Using NLP, pages 127–135, Stroudsburg, PA, USA.
  16. Jack Mostow and Gregory Aist. 2001. Evaluating tutors that listen: An overview of project listen. In K. Forbus andP. Feltovich, editor, Smart Machines in Education: The coming revolution in educational technology., pages 169 – 234. MIT/AAAI Press. find
  17. John Nerbonne, Lauri Karttunen, Elena Paskaleva, Gabor Proszeky, and Tiit Roosmaa. 1997. Reading more into foreign languages. In Proceedings of ANLP’97, pages 135–138. find
  18. Ted Pedersen and Varada Kolhatkar. 2009. Wordnet::senserelate::allwords: a broad coverage word sense tagger that maximizes semantic relatedness. In Proceedings of HLT/NAACL’06, Companion Volume: Demonstration Session, NAACL-Demonstrations ’09, pages 17–20, Stroudsburg, PA, USA. find
  19. S. Petersen. 2007. Natural language processing tools for reading level assessment and text simplification for bilingual education. Ph.D. thesis, University of Washington, Washington, USA. find
  20. Caleb Prichard. 2008. Evaluating l2 readers vocabulary strategies and dictionary use. Reading in a Foreign Language, 20(2):216–231. ↗
  21. D. Radev, T. Allison, S. Blair-Goldensohn, J. Blitzer, A. Çelebi, S. Dimitrov, E. Drabek, A. Hakim, W. Lam, D. Liu, J. Otterbacher, H. Qi, H. Saggion, S. Teufel, M. Topper, A. Winkel, and Z. Zhang. 2004. Mead - a platform for multidocument multilingual text summarization. In Proceedings of LREC 2004, Lisbon, Portugal.
  22. Karthik Raghunathan, Heeyoung Lee, Sudarshan Rangarajan, Nathanael Chambers, Mihai Surdeanu, Dan Jurafsky, and Christopher Manning. 2010. A multi-pass sieve for coreference resolution. In Proceedings of EMNLP’10, pages 492–501.
  23. Advaith Siddharthan. 2004. Syntactic simplification and text cohesion. Technical Report 597, University of Cambridge. find
  24. Dingding Wang, Shenghuo Zhu, Tao Li, Yun Chi, and Yihong Gong. 2008. Integrating clustering and multi-document summarization to improve document understanding. In Proceedings of the 17th ACM conference on Information and knowledge management, CIKM ’08, pages 1435–1436, New York, NY, USA. ACM. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/azab-2013-nlp-based-reading-tool-aiding/