Talk:Natural language generation

Wiki Education Foundation-supported course assignment
This article is or was the subject of a Wiki Education Foundation-supported course assignment. Further details are available on the course page. Student editor(s): Yuxiaosun.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 01:29, 18 January 2022 (UTC)

Hyphen in article name
Currently, the article is titled "Natural-language generation". This spelling (with a hyphen) does not appear to be commonly used outside of this article. For example, the Association for Computational Linguistics' relevant Special Interest Group, SIGGEN, uses the term without a hyphen, as does the Association for Computing Machinery in their CCS classification system (see CCS -> Computing methodologies -> Artificial intelligence -> Natural language processing -> Natural language generation). Similarly, the unhyphenated term seems to be vastly preferred in titles of scientific articles. Should this title be changed? -Ljleppan (talk) 08:14, 27 September 2021 (UTC)


 * I've introduced a formal move request below. -Ljleppan (talk) 09:58, 27 September 2021 (UTC)

Requested move 27 September 2021

 * The following is a closed discussion of a requested move. Please do not modify it. Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a move review after discussing it on the closer's talk page. No further edits should be made to this discussion. 

The result of the move request was: moved. (closed by non-admin page mover) Vpab15 (talk) 18:50, 5 October 2021 (UTC)

Natural-language generation → Natural language generation – The current spelling (with a hyphen) does not appear to be commonly used outside of this article. For example, the Association for Computational Linguistics' relevant Special Interest Group, SIGGEN, uses the term without a hyphen, as does the Association for Computing Machinery in their CCS classification system (see CCS -> Computing methodologies -> Artificial intelligence -> Natural language processing -> Natural language generation). Similarly, the unhyphenated term seems to be vastly preferred in titles of scientific articles, as evidenced by even a cursory search of Google scholar. In fact, I'm not able to find any uses of the hyphenated spelling outside of this Wikipedia article. The related term Natural language processing also uses the unhyphenated form.

NB: There is currently a redirect from Natural language generation to this page, so that needs to be dealt with somehow. Ljleppan (talk) 09:54, 27 September 2021 (UTC)


 * Support. The sources do not appear to hyphenate the term, so neither should Wikipedia.  Note that page was originally created without hyphen before being moved by someone not apparently familiar with the subject matter.  SnowFire (talk) 19:06, 27 September 2021 (UTC)


 * Oppose per WP:HYPHEN #3. (Read it, please.) This article refers to the generation of natural language, not language generation that is natural.  Move the other article instead.  —  AjaxSmack  04:11, 2 October 2021 (UTC)


 * The hyphenated form is not WP:COMMONNAME. Can you provide any reliable sources that use the hyphenated form of the term? -Ljleppan (talk) 07:47, 2 October 2021 (UTC)


 * Strong Support. This is a term, and the usage is crystal clear. WP:HYPHEN is in fact irrelevant – apart from the fact that hyphenation "rules" are not strict and unvarying in English, which is also reflected in WP:HYPHEN. --bonadea contributions talk 09:50, 2 October 2021 (UTC)


 * Support – WP:HYPHEN (guideline) doesn't override WP:COMMONNAME (policy), and the unhyphenated term is what appears in reliable sources. --GoneIn60 (talk) 09:07, 5 October 2021 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
 * Oppose . The proposed move would be very confusing, for people who don't know about this subject, as it would make it look like language generation which is natural, rather than the intended meaning of the generation of natural language. This is precisely the sort of use case that MOS:HYPHEN is designed to address. WP:COMMONNAME is largely irrelevant in this case, as it is a question of how to style an English-language phrase rather than looking at which actual name is dominant. &mdash; Amakuru (talk) 15:20, 5 October 2021 (UTC)
 * It has been pointed out on my talk page that this term isn't intended to be a descriptive phrase of some general process, but a highly specific IT term. As such, and given the evidenced strong preponderance of sourced examples without a hyphen (indeed many uses of the term are in title case), I am amending my !vote to a support. &mdash; Amakuru (talk) 16:12, 5 October 2021 (UTC)

Natural Language Generation is not distinct from machine translation
Contrary to what is suggested in the "Evaluation" section, machine translation is itself a form of conditional natural language generation. Machine translation is very similar to other NLG domains such as text summarization and image captioning, in which an input sequence of data is used by a model to produce unstructured text output. This article should be reorganized to break down sub-problems within natural language generation based on recent surveys in the field (see "Pretrained Language Models for Text Generation: A Survey"). 24.53.246.226 (talk) 12:48, 7 September 2022 (UTC)


 * Defining natural language generation is notoriously difficult. As cited in the lede of the article, Ehud Reiter defines NLG as the subfield of artificial intelligence and computational linguistics that is concerned with the construction of computer systems that can produce understandable texts in English or other human languages from some underlying non-linguistic representation of information. This is mirrored by one of the most comprehensive surveys from the recent years (Gatt and Krahmer, 2018) which uses "NLG" as synonymous for data-to-text generation. A preprint survey by Dong et al. (2021) also defines it as the task of generating text from underlying non-linguistic representation of information. A 2020 article by Robert Dale defines it as task of producing linguistic output from underlying nonlinguistic data. Obviously there is an argument to be made that NLG would encompass all processes that produce natural language output, irrespective of process inputs, but that definition is at minimum controversial. This terminological problem might be worth writing about in itself, but restructuring the whole article seems unnecessary at this point. Ljleppan (talk) 13:28, 7 September 2022 (UTC)

The whole article is outdated
Should it be updated, merged into another article or maybe redirected to a section within another article? DancingPhilosopher ( talk ) 13:56, 4 August 2023 (UTC)