Deep linguistic processing: Difference between revisions

Revision as of 11:41, 4 July 2012

Latest revision as of 09:49, 5 June 2021

(41 intermediate revisions by 21 users not shown)

Line 1:

'''Deep Linguisticlinguistic Processingprocessing''' is a [[natural language processing]] framework which draws on theoretical and [[descriptive linguistics]]. It models language predominantly by way of theoretical syntactic/semantic theory (e.g. [[Combinatory categorial grammar|CCG]], [[HPSG]], [[Lexical functional grammar|LFG]], [[Tree-adjoining grammar|TAG]], the [[Prague School]]). The Deep Linguisticlinguistic Processingprocessing approaches differ from "shallower" methods in that they yield richer, more expressive, and structural representationrepresentations which directly capture [[long-distance dependencies]] or theand underlying [[predicate (grammar)|predicate]]-[[arguementargument]] structure directlystructures.<ref>Timothy Baldwin, Mark Dras, Julia Hockenmaier, Tracy Holloway King, and Gertjan van Noord. 2007. [http://dl.acm.org/citation.cfm?id=1621410.1621415 The impact of deep linguistic processing on parsing technology]. In Proc. of the 10th International Workshop on Parsing Technologies (IWPT-2007), pages 36–8, Prague, Czech Republic.</ref> <br/>

The knowledge-intensive approach of deep linguistic processing requires considerable computational power, and has in the past sometimes been judged as being intractable. However, research in the early 2000s had made considerable advancement in efficiency of deep processing.<ref>Ulrich Callmeier. [http://dl.acm.org/citation.cfm?id=973952.973959 PET – A platform for experimentation with efficient HPSG processing techniques]. Natural Language Engineering, 6(1):99 – 108, 2000.</ref><ref>Hans Uszkoreit. [http://acl.ldc.upenn.edu/coling2002/proceedings/data/area-01/uszkoreit.pdf New Chances for Deep Linguistic Processing] {{Webarchive|url=https://web.archive.org/web/20051103081050/http://acl.ldc.upenn.edu/coling2002/proceedings/data/area-01/uszkoreit.pdf |date=2005-11-03 }}. In Proceedings of COLING 2002, pages xiv–xxvii, Taipei, Taiwan, 2002.</ref> Today, efficiency is no longer a major problem for applications using deep linguistic processing.

==Contrast to "shallow linguistic processing"==

==Deep vs Shallow Linguistic Processing==

Traditionally, deep linguistic processing has been concerned with computational grammar development (for use in both [[parsing]] and generation). These grammargrammars were manually developed, maintained and were computationally expensive to run. In recent years, machine learning approaches (also known as [[shallow linguistic processing]]) have fundamentally altered the field of [[natural language processing]]. The rapid creation of robust and wide-coverage machine learning NLP tools requires substantially lesser amount of manual labor. Thus deep linguistic processing methods have received less attention. However it is the belief of some computational linguists that in order for computers to understand natural language or [[inference]], detailed syntactic and semantic representation is necessary.

Moreover, shallow methods does lack human language 'understanding'; While humans can easily understand a sentence and it's meaning, shallow linguistic processing might lack human language 'understanding'. For example<ref>U. Schafer. 2007. ¨ [http://scidok.sulb.uni-saarland.de/volltexte/2007/1326/pdf/Dissertation_1383_Schae_Ulri_2007.pdf Integrating Deep and Shallow Natural Language Processing Components – Representations and Hybrid Architectures]. Ph.D. thesis, Faculty of Mathematics and Computer Science, Saarland University, Saarbrucken, Germany.</ref>:

<br>

a) Things would be different if Microsoft was located in Georgia.

b) The National Institue for Psychology in Israel was established in May 1971 as the Israel Center for Psychobiology by Prof. Joel.

<br>

In sentence A, a shallow [[information extraction]] system might infer wronlgy that Microsoft's headquarters was located in Georgia. While as humans, we understand from the sentence that Mircosoft office was never in Georgia.

MoreoverHowever, shallowit methodsis doesthe lackbelief humanof some computational linguists{{Who|date=August 2012}} that in order for computers to understand natural language 'understanding';or [[inference]], detailed syntactic and [[Semantic analysis (knowledge representation)|semantic representation]] is necessary. Moreover, Whilewhile humans can easily understand a sentence and it'sits meaning, shallow linguistic processing might lack human language 'understanding'. For example:<ref>U. Schafer. 2007. ¨ [http://scidok.sulb.uni-saarland.de/volltexte/2007/1326/pdf/Dissertation_1383_Schae_Ulri_2007.pdf Integrating Deep and Shallow Natural Language Processing Components – Representations and Hybrid Architectures]. Ph.D. thesis, Faculty of Mathematics and Computer Science, Saarland University, Saarbrucken, Germany.</ref>: <br/>

:a) ''Things would be different if Microsoft waswere located in Georgia.''

In sentence A(a), a shallow [[information extraction]] system might infer wronlgywrongly that Microsoft's headquarters was located in Georgia. While as humans, we understand from the sentence that MircosoftMicrosoft office was never in Georgia.<br/>

:b) ''The National InstitueInstitute for Psychology in Israel was established in May 1971 as the Israel Center for Psychobiology by Prof. Joel.''

In sentence (b), a shallow system could wrongly infer that Israel was established in May 1971. Humans know that it is the National Institute for Psychobiology that was established in 1971.<br/>

In summary of the comparison between deep and shallow language processing, deep linguistic processing provides a knowledge-rich analysis of language through manually developed grammars and language resources. Whereas, shallow linguistic processing provides a knowledge-lean analysis of language through statistical/machine learning manipulation of texts and/or [[Annotation|annotated linguistic]] resource.

==Sub-communities==

"Deep" computational linguists are divided in different sub-communities based on the grammatical formalism they adopted for deep linguistic processing. The major sub-communities includes the:

*'''DE'''ep '''L'''inguistic '''P'''rocessing with '''H'''PSG - '''IN'''itiative ([[DELPH-IN]]) collaboration working with the [[HPSG]] formalism. The [http://cslipublications.stanford.edu/HPSG/ HPSG Conference] is the central conference to share knowledge/advancement of [[HPSG]] based deep processing.

*[https://pargram.b.uib.no/ '''ParGram'''/'''ParSem'''] is international collaboration on [[Lexical functional grammar|LFG]]-based grammar and semantics development. The [http://www.essex.ac.uk/linguistics/external/lfg/FAQ/conferences.html LFG Conference] is the central conference to share knowledge/advancement of [[Lexical functional grammar|LFG]] based deep processing.

*XTAG Research group working with the TAG formalism. The [http://alpage.inria.fr/tagplus11/doku.php?id=start TAG+ conference] is the central conference to share knowledge/advancement of [[Tree-adjoining grammar|TAG]] based deep processing.

The shortlist above is not exhaustively representative of all the communities working on deep linguistic processing.

==See also==

*[[Combinatory categorial grammar]]

*[[Head-driven phrase structure grammar]]

*[[Lexical functional grammar]]

*[[Natural language processing]]

*[[Tree-adjoining grammar]]

==References==

{{Reflist}}

{{Natural Language Processing}}

{{Compu-stub}}

{{Ling-stub}}

[[Category:Natural language processing]]