Wikifact

This is a proposal for a new WMF sister project.
Wikifact
Status of the proposal
Statusunder discussion
Details of the proposal
Project descriptionWikifact is a resource for real-time collaborative fact checking.
Is it a multilingual wiki?To be determined.
Potential number of languagesMultiple languages.
Proposed taglineA resource for real-time collaborative fact checking.
New features to requireTo be determined.

IntroductionEdit

This is a proposal for Wikifact, a resource for real-time collaborative fact checking.

A new Wikifact project could serve as a fact-checking resource for the editors of Wikinews, Wikipedia, and Wikisource articles.

A new Wikifact project could also serve as a fact-checking resource for a broader set of end-users. Through an API, end-users could perform real-time fact checking via Wikifact while authoring or reviewing documents.

Proposed byEdit

Alternative namesEdit

  • Wikifacts
  • Wikifactchecking

Related projects / proposalsEdit

See also: https://www.wikicred.org/

Mailing list linksEdit

DemosEdit

None.

ScenariosEdit

In this section, we describe some Wikifact scenarios.

Fact-checking speeches and statementsEdit

An important scenario for Wikifact is the fact-checking of political speeches and statements. This could be done with a combination of Wikisource and Wikifact.

Editors could create new Wikisource articles for political speeches and, while transcribing them and adding multimedia content, make use of new templates to indicate specific content for synchronization with Wikifact articles and related crowdsourced fact-checking processes. That is, editors could collaboratively seek to validate and verify any claims made in political speeches or statements, utilizing one or more new wiki templates to synchronize their content with Wikifact articles about individual statements, claims, or facts.

Each Wikifact article would contain a number of fact-checking annotations about its particular statements, claims, or facts.

Fact-checking social mediaEdit

Wikifact could also be utilized for fact-checking social media content, statements made on microblogging and sites such as Twitter and Facebook.

Fact-checking Wikinews, Wikipedia, or Wikisource articlesEdit

Editors could make use of one or more new templates to indicate or to select statements, claims, or facts for synchronization with Wikifact.

Envisioned features include the capability to navigate from statements, claims, or facts in Wikinews, Wikipedia, or Wikisource articles to Wikifact articles about them. Envisioned features include the capability to also navigate to ad-hoc collections of statements, claims, or facts (such as those co-occurring in Wikinews, Wikipedia, or Wikisource articles).

Both editors and end-users could hover over content in template spans to view real-time Wikifact information in tooltips and could click on content in template spans – or on superscript hyperlink symbols after such spans – to navigate to the content’s corresponding Wikifact article. Real-time information from multiple template spans in an article could also be aggregated, collected together, and presented to end-users. One possibility for this visualization involves categorizing the annotations into informational messages, warnings, and errors, in which case aggregations from multiple template spans could be presented in toggleable table views, resembling software development IDE’s.

Also, whenever a statement, claim, or fact were updated or annotated via Wikifact, or as events were to unfold on Wikifact, editors of any dependent Wikinews, Wikipedia, or Wikisource articles could be notified so as to be able to revisit their articles and/or to contribute on Wikifact.

Fact-checking while authoring or reviewing documentsEdit

Wikifact could, by means of an API, serve as a resource for fact-checking while end-users author documents in word processors and while they review documents in Web browsers. In an annotational approach, end-users could make use of Wikifact and related services to view fact-checking annotations for content selected in documents.

Technical discussion topicsEdit

Wikidata as a backendEdit

With schemas for statements, claims, or facts, schemas which define their structure and interrelations, Wikifact could utilize Wikidata as a backend.

New wiki templatesEdit

New wiki templates are possible for indicating or selecting content for synchronization with Wikifact.

{{statement|User content goes here.}}

or

{{claim|User content goes here.}}

or

{{fact|User content goes here.}}

Both editors and end-users could hover over content in template spans to view real-time Wikifact information in tooltips and could click on content in template spans – or on superscript hyperlink symbols after such spans – to navigate to the content’s corresponding Wikifact article. Real-time information from multiple template spans in an article could also be aggregated, collected together, and presented to end-users. One possibility for this visualization involves categorizing the annotations into informational messages, warnings, and errors, in which case aggregations from multiple template spans could be presented in toggleable table views, resembling software development IDE’s.

The use of such wiki templates in Wikinews, Wikipedia, or Wikisource articles could, potentially, automatically create new Wikifact articles and/or synchronize content with existing Wikifact articles.

Indicating statements, claims, or facts upon which articles depend could also be done via document metadata.

EpistemologyEdit

AnnotationEdit

One can view ascribing Boolean or other truth values to statements, claims, or facts as a form of annotation.

Models of annotation include the W3C Web Annotation Data Model.

Schema.org’s ClaimReview presently annotates statements with numerical values from an interval, e.g. x out of N units.

Fact-checking annotationsEdit

Annotations utilized by fact-checkers include, but are not limited to: “true”, “mostly true”, “half true”, “partially true”, “mostly false”, “false”, “disputed”, “misleading”, “contains omissions”, “contains exaggerations”, “contains distortions”, “contains hearsay”, and so forth. The approaches under discussion can support all of these fact-checking annotations.

Perhaps fact-checking annotations could be categorized. One approach for such categorization includes the use of informational messages, warnings, and errors. Benefits of such approaches would include that annotations could be easily integrated, merged together, from multiple sources or service providers and that annotations could be easily organized and viewed by end-users.

Multi-statement annotationsEdit

Annotations could annotate multiple passages of text from one or more documents.

See also: Web Annotation Data Model § Cardinality of Bodies and Targets, Web Annotation Data Model § Sets of Bodies and Targets.

IndeterminacyEdit

Unannotated statements, claims, or facts can be described as being “indeterminate”.

ParaconsistencyEdit

It may occur that annotations on Wikifact are paraconsistent. Some statements may be disputed, annotated, by some parties, as “true”, and by other parties, as “false”. In an annotational approach, statements could be annotated as being “disputed” without paradox or error.

Quotations and attestationsEdit

Quotations and attestations are useful to consider for Wikifact.

Schema.org’s quotation and Wikidata’s quotation could be of use.

Citations and referencesEdit

Annotations of statements, claims, or facts could cite and quote from referenced materials.

ExplanationEdit

Annotations of statements, claims, or facts could provide text, hypertext, or other structured content which explains the annotations.

ArgumentationEdit

Annotations of statements, claims, or facts could also provide machine-utilizable argumentation, justifications, or rationales. Annotations could themselves be annotated.

Scientific models and theoriesEdit

If Wikifact is desired to facilitate advanced scenarios for the sciences, we might want to consider scientific models or theories when formalizing schemas for statements, claims, or facts. That is, scientific models or theories could be relevant when defining the structure of and interrelations between statements, claims, or facts.

A statement could be annotated as “true” for one model or theory, Newtonian physics, but annotated as otherwise for a different model or theory, Einsteinian physics. In such a case, it might be desired that, instead of the statement being “disputed”, the statement could be described as “true” for a particular scientific model or theory.

Artificial intelligenceEdit

Annotations could be generated by human editors and software tools.

Software tools could process sets of fact-checking annotations – some explained, others argued for – and decide whether a given statement were true, false, or indeterminate.

Alerts and notificationsEdit

Whenever a statement, claim, or fact were updated or annotated via the Wikifact website, editors of any dependent Wikinews, Wikipedia, or Wikisource articles could be notified so as to be able to revisit their articles or to contribute on Wikifact.

Editors might opt to receive individual email alerts, to receive digests or automatically-generated newsletters which contain changes or updates relevant to their interests which occurred over a period of time. Editors could also review dynamic, personalized, automatically-generated pages of content when they login to or visit Wikinews, Wikipedia, or Wikisource.

Wikifact alerts and notifications could be interoperable with the Echo alerts and notifications system.

Communication channels in addition to or instead of email (e.g. instant messaging applications, collaboration and productivity software) could also be opted into for Wikifact alerts and notifications.

Statements and collections of statementsEdit

Facts can change and facts can unfold.

Facts can change. An edit of or an annotation of a statement, claim, or fact could result in alerting or notifying interested editors.

Facts can unfold. New statements, claims, or facts could be added to collections. Whenever a new statement, claim, or fact were added to a collection or sequence, each previous element of the collection or sequence could be pinged as a result, this alerting or notifying any interested editors. Such collections or sequences of statements, claims, or facts could be called “timelines”, “threads”, or “stories”.

Statements, claims, or facts could also be collected together into ad-hoc collections such as those co-occurring in articles, speeches, or reports.

Folksonomic keywords, categories, or hashtags could be utilized in wiki content for individual statements, claims, or facts.

ParaphrasesEdit

See also: Wikipragmatica

Individual statements, facts, or claims could map to paraphrase clusters – representing abstract, phrasing-invariant, statements – and these clusters could map to wiki-based fact-checking articles. In the model presently envisioned, editors could annotate and argue about clusters as well as redirects. Most re-phrasing or interpretive redirects will be uncontroversial but, for some cases, re-phrasings or interpretations would be the locus of some debate. For such scenarios, one could provide multiple options and utilize an HTTP response code of 300.

Each statement, claim, or fact could have a URL, for instance https://www.wikifact.org/statements/33DCF305-3A4D-4024-9AD7-CCB1A29054E2, and each paraphrase cluster could have a URL, for instance https://www.wikifact.org/clusters/D006871E-24A6-428F-BD1F-D20C3C7B7685. The URL for an individual statement, claim, or fact could redirect to a URL for the paraphrase cluster which contains it. This could convenience processes of semi-automated, collaborative paraphrasing; that is, in the event of an erroneous paraphrasing, editors or software tools could edit a redirect page to re-cluster an individual statement, claim, or fact to an updated cluster of paraphrases. At the URL for a paraphrase cluster would be a human-editable sequence of annotations and arguments about a statement, claim, or fact.

The emergent feature of URL-addressability could facilitate and convenience Web-based communication about statements, claims, and facts, their phrasing-invariant clusters, and wiki-based fact-checking articles.

Associated metadata could be encoded in JSON-LD to adhere to the latest standards.

Statement patternsEdit

Statement patterns could be expressed and these patterns could be utilized via URL query strings. Nouns or noun phrases could be provided as arguments. That is, arguments for thematic relations could be provided utilizing Wikidata lexemes and entities. For example, https://www.wikifact.org/patterns/293FCD5D-27A7-498A-81C3-C78EF0F9D9A2?agent=Q42&patient=Q89 could represent a set of statements expressing that “Douglas Adams ate an apple.”

Linguistic contextsEdit

With respect to schemas for statements, claims, or facts, schemas defining the structure of and relationships between them, one can consider that statements, claims, or facts could reference, or otherwise be attached to, their linguistic contexts. Linguistic contexts could be paragraphs, passages of text, or entire documents. Referencing the contexts of statements, claims, or facts could be useful for natural language processing, anaphora resolution, natural language understanding, and machine-learning algorithms.

Conveniently, for wiki-based content, snapshots of articles can be referenced by URL’s, e.g. https://en.wikipedia.org/w/index.php?title=Context_(language_use)&oldid=983879664, and, for other domains, one could use a Web archive.

Misinformation is rarely uttered as a standalone statement or claim. It is critical that we not only identify the misinformation or disinformation, but that we also understand the context in which it was communicated. With this knowledge, we can enhance and automate detection.

Semi-automated and automated content generationEdit

Could Wikifact content be generated by the processing of existing Wikinews, Wikipedia, and Wikisource articles? By parsing existing Wikinews, Wikipedia, and Wikisource articles, focusing on sentences which cite references, could a Wikifact resource be bootstrapped? Could software tools be created to process existing articles to produce initial Wikifact content?

A real-time fact-checking resourceEdit

Through an API, end-users could perform real-time fact checking via Wikifact while authoring or reviewing documents. In an annotational approach, end-users could make use of Wikifact and related services to view fact-checking annotations for content selected in documents.

People interestedEdit

  1.   Support. --DougClark55 (talk) 17:50, 18 February 2021 (UTC)
  2.   Support:Much neededVis M (talk) 12:12, 25 February 2021 (UTC)
  3.   Support. It will be interesting to have the project as a structured edition of Wikinews, similarly as Structured Commons Project. --Csisc (talk) 13:57, 22 March 2021 (UTC)

People opposedEdit

  1.    Oppose - Hérisson grognon 13:59, 14 May 2021 (UTC)