Talk:Sentiment analysis

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. Peer reviewers: Forever kka.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 08:58, 17 January 2022 (UTC)

Wiki Education Foundation-supported course assignment

This article was the subject of a Wiki Education Foundation-supported course assignment, between 2 September 2020 and 9 December 2020. Further details are available on the course page. Student editor(s): Thetrailblazer.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 08:58, 17 January 2022 (UTC)

Needs reworking

This page retains its incremental authoring history and needs to be pulled into shape more systematically. Jussi Karlgren (talk) 08:50, 30 May 2017 (UTC)

Types of sentiment analysis

The passage on sentiment analysis that explains how scale based approaches work has a minor ambiguity. Does it refer to techniques that assess the sentiment of a given term by adjusting the score relative to the other terms in the environment or does it refer to techniques that assess the sentiment of sentences by pooling the scores of different words? I am going to assume that it refers to the former and adjust the entry slightly --Rilinger (talk) 15:59, 12 May 2016 (UTC)

The polarity sentiment analysis section should have its own subheading under Types in the table of contents. I was learning about sentiment analysis, and wanted to look up what the polarity metric is. This article made the first impression of not covering the polarity metric, as it was not listed on the table of contents. -- Uneven-sunflower (talk) 20:22, 29 December 2024 (UTC)

Untitled

In reviewing this page, it seems that the weakest part of the page is the last section attempting to associate web2.0 and sentiment analysis. The rest could use some technical cleanup, but that section doesn't seem to add a lot of value. Anybody feel differently? Sredmore (talk) 06:22, 22 October 2010 (UTC)

A section on the ethics of sentiment analysis would be more than welcome.

I think this needs a reference "The shorter the string of text, the harder it becomes." because I would have thought that the shorter the text the EASIER it would be to determine sentiment, due to the sentiment being more condensed. Lanksalot

I also have doubts about "The shorter the string...". Yesterday I heard a presentation from an expert on sentiment analysis who said that shorter texts are easier. MikeStantonBcn (talk) 07:51, 11 February 2011 (UTC)

Last year i benefited from this page greatly, I wanted to contribute a bit. Sentiment analysis have shifted to microblogging tools like twitter. Analysis of blogs, articles or news pieces are still there, but many industrial applications have been built on top of Twitter. I added a reference to a recent work(EDIT: THIS IS MY PAPER, I WANTED TO ADD THIS INFORMATION IN CASE YOU USE MY ADDED REFERENCE). I will look for others as well, but I think Twitter part of this Wikipedia article is very inadequate. This was the article I added: http://portal.acm.org/citation.cfm?id=1964867 —Preceding unsigned comment added by 193.206.170.151 (talk) 13:36, 15 April 2011 (UTC)

Evaluation

Section on Evaluation needs to mention some references!!! It is too factual in order to be unsuported by literature. Otherwise this article looks good. —Preceding unsigned comment added by 158.125.224.201 (talk) 08:26, 28 April 2011 (UTC)

Agreed. I'm trying to find a citation for this claim in the lead sentence of the Evaluation section: "The accuracy of a sentiment analysis system is, in principle, how well it agrees with human judgments. This is usually measured by precision and recall. However, human raters typically agree about 70% of the time" barryparr —Preceding undated comment added 01:11, 17 November 2011 (UTC).

Added some recent references, but needs much more. Jussi Karlgren (talk) 08:49, 30 May 2017 (UTC)

Typo

This sentence: "Automation impacts approximately 23% of comments that are correctly classified by humans." I am really not sure what the author means by "impacts." My guess is that this was intended to say "Automated methods correctly label approximately 23% of comments that are correctly classified by humans, where correct human classification is estimated by inter-rater agreement." Rpgoldman (talk) 00:26, 5 March 2019 (UTC)

Concerns over article content: Intensity ranking

Hello!

On the 21th of October I added a new sub-heading under "Types" labelled 'Intensity Ranking'. And I proceeded to describe what that function does. But that edit was removed by MrOllie. Was wondering what did I do wrong or how I could improve upon it?

Thanks for the guidance!

Turq6ise (talk) 07:30, 5 November 2021 (UTC)

As I said last time you asked me about this on my talk page, contributions to Wikipedia must not promote individual companies or services such as SenticNet. Sourcing should come from independent secondary sources, preferably review articles in this space. MrOllie (talk) 11:47, 5 November 2021 (UTC)

Hi Mr Ollie, I recently edited the article and removed promotion of individual companies/services. Sourcing came from independent secondary sources! Do let me know if there is any improvements I can make!

Thank you. Turq6ise (talk) 07:31, 19 November 2021 (UTC)

Proposed addition: Literary and narrative analysis subsection

Disclosure: As noted on my user page (User:J2000ai#Conflict of interest), I have a conflict of interest regarding the Sentiment analysis article as co-developer of SentimentArcs and co-author (with Katherine Elkins) of related publications, including The Shapes of Stories: Sentiment Analysis for Narrative (Cambridge University Press, 2022). I will not edit the article directly and am proposing this via talk page per WP:COI.

I suggest adding the following new subsection to cover literary and narrative uses of sentiment analysis, an emerging application area documented in peer-reviewed sources and mainstream coverage (e.g., The Atlantic, MIT Technology Review). This would fit well as a new ==== Literary and narrative analysis ==== level-4 subsection, potentially under an existing "Applications"-related area (e.g., expanding from "Application in recommender systems" or as a standalone if no broad "Applications" section exists yet; the current article structure has application-focused content in sections like Web 2.0 and recommender systems, so this complements them by addressing computational literary criticism).

Proposed text (to insert as ==== Literary and narrative analysis ====):

Literary and narrative analysis

Sentiment analysis has been applied to literary studies to trace how emotional valence changes across the arc of a narrative, rather than classifying a document's overall polarity. Reagan et al. (2016) used sentiment analysis on over 1,300 Project Gutenberg novels to identify six core emotional trajectories that recur across fiction, a finding covered in The Atlantic and MIT Technology Review. Beginning in 2019, Katherine Elkins and Jon Chun applied diachronic sentiment analysis to literary texts using a hybrid computational and close reading approach, demonstrating that novels traditionally considered plotless exhibit underlying emotional structures. Chun developed SentimentArcs, the first comprehensive ensemble framework for diachronic sentiment analysis, combining over three dozen sentiment models, and Elkins developed the methodology for applying SentimentArcs to literature in The Shapes of Stories: Sentiment Analysis for Narrative (Cambridge University Press, 2022). The methodology has since been extended to political speeches, film, social media narratives, and fairy tales. A 2023 critical survey in Digital Humanities Quarterly examined the growing use of sentiment analysis tools in literary studies, evaluating six representative methods and their applicability to literary texts.

Thanks for considering this addition. J2000ai (talk) 15:37, 23 March 2026 (UTC) J2000ai (talk) 15:37, 23 March 2026 (UTC)