Transcript Slides
NLP CW1–09 Text simplification 01/13 Can text simplification help a reader with a low to basic level of language? By Barney Staddon NLP CW1–09 Text simplification • The need for text simplification • The approaches • The problems • The solutions (or attempts to solve problems!) • Conclusions 02/13 NLP CW1–09 Text simplification 03/13 The need for text simplification “If complex texts can be made simpler, sentences become easier to process, both for programs and humans” Chandrasekar et al (1996) NLP CW1–09 Text simplification 04/13 The need for text simplification Natural Language Processing: Machine translation Simplification Text summarization Text analysis NLP CW1–09 Text simplification The need for text simplification Human Processing: Greater access to information • English only spoken as first language by 4.7% (CIA, 2009) • 18% of world population are illiterate (CIA, 2009) • Newspapers, official information, subtitles, print or online. Language education • Manual simplification is expensive and time consuming Readers with disabilities • Aphasia 05/13 NLP CW1–09 Text simplification The approaches Author-validated simplification: 06/13 NLP CW1–09 Text simplification The approaches Post-authored simplification: 07/13 NLP CW1–09 Text simplification 08/13 The problems INPUT Syntactic analysis: What is the structure? Syntactic simplification: Can it be made simpler? Lexical simplification: Is the vocabulary difficult? Cohesion analysis: Does it still make sense? OUTPUT NLP CW1–09 Text simplification The solutions Parsers and lemmatizers Sentence splitting & pronoun resolution Machine learning approach Rules based approach Lexical databases Author validation 09/13 NLP CW1–09 Text simplification 10/13 John gets the bus to university but he always gets there incredibly late even if he leaves early. John gets the bus to university. However, John always gets to university incredibly late. John leaves early. John gets the bus to university. However, John always gets to university very late. John leaves early. John gets the bus to university. John leaves early. However, John always gets to university very late. Syntactic simplification Lexical simplification Cohesion analysis NLP CW1–09 Text simplification 11/13 Conclusions Post-authored simplification - Error prone, can lack cohesion - Relies on rules, (judgements based on experience?) Author-validated simplification - Very slow, requires author - Side steps problems of semantic meaning - Works well, ensures cohesion Can text simplification help a reader with a low to basic level of language? Yes, but currently without main benefits of computerization! NLP CW1–09 Text simplification (Source: http://www.storyscribe.com/sssoftware/stylewriter-images/screenshots/SS12.gif) 12/13 NLP CW1–09 Text simplification 13/13 Can text simplification help a reader with a low to basic level of language? Questions References: Chandrasekar R., Doran, C. and Srinivas, B. (1996). Motivations and Methods for Text Simplification. Proceedings of the 16th conference on Computational linguistics, (2). pp. 1041 – 1044. [Online]. Available from: http://www.aclweb.org/anthology/C/C96/C96-2183.pdf [Accessed 13/11/09] Central Intelligence Agency. (2009). The World Factbook. [Online]. Available from: https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html [Accessed 13/11/09]