Latest / Marketing Automation in 2026 / Building Author Hubs: ORCID, Crossref, and Scholar Profiles as Trust Primitives

Transcript

Building Author Hubs: ORCID, Crossref, and Scholar Profiles as Trust Primitives · Marketing Automation in 2026

• 0:00
  SPEAKER_00: Building Author Hubs, Worcid, Crossref, and Scholar profiles as trust primitives.
• 0:05
  SPEAKER_00: Author hubs are public connected records that help people and machines answer three basic questions.
• 0:11
  SPEAKER_00: Who created this work?
• 0:12
  SPEAKER_00: What exactly is the work?
• 0:14
  SPEAKER_00: Why should anyone trust the connection between the person, the work, and the organization?
• 0:19
  SPEAKER_00: For human readers, an author hub may look like a profile page with a biography, publications, credentials, and links.
• 0:26
  SPEAKER_00: For artificial intelligence systems, it is more useful to think of it as a linked evidence graph.
• 0:31
  SPEAKER_00: A strong author hub connects a person to an ORCID ID, a publication, dataset, report, or software package to a digital object identifier, a work to complete Crossref or data site metadata, a researcher to verified affiliations and organizations, a public profile to a stable institutional web page, contributions to clear roles such as those in the contributor roles, taxonomy, current and past roles to dates, sources, and update records.
• 0:58
  SPEAKER_00: These connections can improve identity resolution, retrieval, citation accuracy, and source verification.
• 1:05
  SPEAKER_00: They do not guarantee that an artificial intelligence system will cite an author.
• 1:09
  SPEAKER_00: There is no credible public evidence that adding one ORCID ID produces a fixed increase in artificial intelligence citations.
• 1:16
  SPEAKER_00: The stronger claim is narrower and more defensible.
• 1:19
  SPEAKER_00: Good identity and work metadata make it easier for search and research systems to find, match, verify, and correctly cite scholarship.
• 1:28
  SPEAKER_00: This article reflects public documentation and research available as of August 13, 2026.
• 1:34
  SPEAKER_00: Why Author Hubs Matter to Artificial Intelligence Systems?
• 1:37
  SPEAKER_00: Artificial intelligence systems do not all use the same search indexes, databases, or ranking methods.
• 1:43
  SPEAKER_00: Some rely on live web search.
• 1:45
  SPEAKER_00: Others use scholarly databases, licensed indexes, internal knowledge stores, or retrieval systems that search documents before generating an answer.
• 1:54
  SPEAKER_00: Still, many systems face the same technical problems.
• 1:57
  SPEAKER_00: Names may be spelled in different ways.
• 1:59
  SPEAKER_00: Several researchers may share the same name.
• 2:02
  SPEAKER_00: A work may have a preprint, conference version, accepted manuscript, and final version.
• 2:07
  SPEAKER_00: A journal article may be linked to a dataset, software package, grant, or correction.
• 2:12
  SPEAKER_00: An author may change institutions or names.
• 2:14
  SPEAKER_00: A web page may describe a credential without showing who issued it or when it was valid.
• 2:19
  SPEAKER_00: Persistent identifiers and structured metadata help solve these problems.
• 2:23
  SPEAKER_00: The digital object identifier system describes an identifier as a persistent name for a digital, physical, or abstract object.
• 2:31
  SPEAKER_00: Its value depends not only on the identifier, but also on the metadata connected to it.
• 2:36
  SPEAKER_00: DOI Handbook.
• 2:37
  SPEAKER_00: A useful way to think about an author hub is as four connected layers.
• 2:41
  SPEAKER_00: For the full table, please open this article on autopod.co.
• 2:45
  SPEAKER_00: The goal is not to create a profile full of impressive words.
• 2:49
  SPEAKER_00: The goal is to create a record in which important claims can be checked.
• 2:54
  SPEAKER_00: The three main parts of an author hub ORCID.
• 2:57
  SPEAKER_00: The person identity layer.
• 2:59
  SPEAKER_00: ORCID stands for Open Researcher and Contributor ID.
• 3:03
  SPEAKER_00: It provides a persistent identifier for a person and helps distinguish researchers who have similar names, use different nameforms, or change their names over time.
• 3:12
  SPEAKER_00: ORCID Overview.
• 3:14
  SPEAKER_00: An ORCID record may contain names and name variants, employment and education, research works, funding, peer review activity, editorial service, memberships and distinctions, research resources, links to other identifiers.
• 3:29
  SPEAKER_00: The most important feature is not the 16-digit number itself, it is the network of supported connections attached to that number.
• 3:36
  SPEAKER_00: For example, authenticated connections are stronger.
• 3:39
  SPEAKER_00: Organizations should collect ORCID IDs through an authenticated sign-in process, rather than asking people to type an identifier into a form.
• 3:47
  SPEAKER_00: ORCID recommends using its authorization process so the researcher confirms the correct record.
• 3:53
  SPEAKER_00: ORCID integration and application programming interface guidance.
• 3:56
  SPEAKER_00: This distinction matters.
• 3:58
  SPEAKER_00: Unverified claim, this ORCID ID belongs to Jane Smith.
• 4:02
  SPEAKER_00: Authenticated claim, Jane Smith signed in through ORCID and authorized this system to use her ID.
• 4:08
  SPEAKER_00: Institutional assertion.
• 4:10
  SPEAKER_00: The university confirms that this person held this affiliation during these dates.
• 4:14
  SPEAKER_00: ORCID also stores the source and origin of assertions.
• 4:18
  SPEAKER_00: A work or affiliation may be added by the researcher, a publisher, a funder, or another trusted organization.
• 4:24
  SPEAKER_00: ORCID Trust Guidance.
• 4:26
  SPEAKER_00: This creates a useful trust signal for both people and machines.
• 4:30
  SPEAKER_00: The record does not only say what is claimed, it can also show who made the claim and when it was added.
• 4:36
  SPEAKER_00: Important limitation An ORCID ID is not a guarantee that every item on a record is correct.
• 4:42
  SPEAKER_00: ORCID itself encourages users to examine trust markers, sources, dates, and the organizations that added information.
• 4:49
  SPEAKER_00: ORCID trust markers.
• 4:50
  SPEAKER_00: Therefore, organizations should not treat has an ORCID ID as equal to, is fully verified.
• 4:56
  SPEAKER_00: A better model is self-asserted, authenticated, institutionally confirmed, publisher confirmed, funder confirmed, recently reviewed, crossref and datasite.
• 5:07
  SPEAKER_00: The work identity layer.
• 5:08
  SPEAKER_00: ORCID identifies the person.
• 5:10
  SPEAKER_00: Crossref and data site help identify the work.
• 5:13
  SPEAKER_00: A digital object identifier can identify journal articles, books and book chapters, conference papers, datasets, software, reports, preprints, peer reviews, standards, posters and presentations.
• 5:26
  SPEAKER_00: A DOI should resolve to a stable landing page.
• 5:29
  SPEAKER_00: More importantly, it should have complete metadata describing the work.
• 5:33
  SPEAKER_00: Crossref recommends recording information such as family name and given name, contributor roles, affiliations, organization identifiers, ORCID IDs, publication dates, abstracts, references, funding, licenses, version information, updates and corrections, Crossref Required and Recommended Metadata.
• 5:54
  SPEAKER_00: Crossref describes its metadata as a shared resource that can be accessed through web interfaces, application programming interfaces, and bulk downloads.
• 6:04
  SPEAKER_00: Downstream systems can use that information to discover and connect scholarly outputs.
• 6:09
  SPEAKER_00: Crossref metadata retrieval and datasite serves a similar function for datasets and many other research outputs.
• 6:16
  SPEAKER_00: Its current metadata schema 4.7, released on March 3rd, 2026, supports creators, contributors, ORCID identifiers, affiliations, organization identifiers, related identifiers, funding, descriptions, and resource types.
• 6:33
  SPEAKER_00: DataSite metadata schema supports connecting creator and contributor records to ORCID identifiers and organizational affiliations to organization identifiers.
• 6:42
  SPEAKER_00: A DOI is not a quality badge.
• 6:45
  SPEAKER_00: A DOI proves that a record exists in a DOI registration system.
• 6:49
  SPEAKER_00: It does not prove that the research is correct, the publication is peer-reviewed, the journal is reputable, the author is an expert, the findings are important.
• 6:57
  SPEAKER_00: Crossref makes this point clearly.
• 7:00
  SPEAKER_00: The DOI itself does not signify the value or accuracy of the object.
• 7:04
  SPEAKER_00: The surrounding metadata provides context and enables connections.
• 7:08
  SPEAKER_00: This is why a complete DOI record is much more useful than a DOI string placed at the bottom of a page.
• 7:14
  SPEAKER_00: Contributor roles add meaning.
• 7:16
  SPEAKER_00: In July 2026, Crossref Schema 5.5 added better support for multiple contributor roles, corresponding authors, and the contributor roles taxonomy.
• 7:26
  SPEAKER_00: The contributor roles taxonomy contains 14 roles, including conceptualization, data curation, formal analysis, investigation, methodology, software, supervision, validation, visualization, writing the original draft, writing, review, and editing.
• 7:44
  SPEAKER_00: The taxonomy was designed to make contributions more transparent and support credit, accountability, research assessment, and research integrity.
• 7:52
  SPEAKER_00: For artificial intelligence systems, this creates a richer answer than Jane Smith is an author.
• 7:58
  SPEAKER_00: A system can potentially distinguish among the person who designed the study, the person who wrote the software, the person who analyzed the data, the person who supervised the project, the person who wrote the original manuscript.
• 8:10
  SPEAKER_00: That is especially important for large research teams and technical projects.
• 8:15
  SPEAKER_00: Scholar profiles make the identity graph understandable to people and easier for search systems to interpret.
• 8:21
  SPEAKER_00: A robust profile should include a stable institutional page, a consistent name, name variants where appropriate, ORCID ID, current and past affiliations, areas of research, selected works with DOI links, datasets, software, and reports, funding and awards, editorial or review roles, contributor roles, dates for current and past positions, a last review date, and links to other public profiles.
• 8:46
  SPEAKER_00: Google Scholar profiles allow researchers to list publications, make the profile public, see who cites their work, and track citation metrics.
• 8:55
  SPEAKER_00: A public profile with a verified institutional email address may be eligible to appear in Google Scholar search results.
• 9:01
  SPEAKER_00: However, Google Scholar should be treated as a visibility and discovery layer, not the main source of truth.
• 9:07
  SPEAKER_00: Researchers can manage their own publication lists, and automatic matching can include incorrect or duplicate works.
• 9:14
  SPEAKER_00: Citation counts also vary across databases.
• 9:17
  SPEAKER_00: OpenALEX offers another useful discovery layer.
• 9:20
  SPEAKER_00: It maintains disambiguated author records and connects authors, works, institutions, sources, and topics.
• 9:27
  SPEAKER_00: OpenALEX states that it draws information from sources such as Crossref and DataSite, then matches authors to ORCID IDs and affiliations to organization identifiers.
• 9:37
  SPEAKER_00: The best practice is therefore profile federation.
• 9:40
  SPEAKER_00: No single profile needs to contain everything.
• 9:42
  SPEAKER_00: They should point to one another and use the same identifiers.
• 9:46
  SPEAKER_00: How these signals can affect artificial intelligence citations.
• 9:49
  SPEAKER_00: It is important to separate five different events.
• 9:52
  SPEAKER_00: Discovery.
• 9:53
  SPEAKER_00: Can a system find the page or work?
• 9:55
  SPEAKER_00: Identity resolution.
• 9:57
  SPEAKER_00: Can it tell which person and organization are involved?
• 9:59
  SPEAKER_00: Retrieval.
• 10:00
  SPEAKER_00: Does the system bring the work into its evidence set?
• 10:03
  SPEAKER_00: Citation selection.
• 10:04
  SPEAKER_00: Does it cite the work in the answer?
• 10:06
  SPEAKER_00: Citation correctness.
• 10:08
  SPEAKER_00: Does the citation point to the right work and support the claim?
• 10:11
  SPEAKER_00: ORCID, DOI metadata, and scholar profiles can help most directly with the first three steps.
• 10:17
  SPEAKER_00: They may also improve citation correctness.
• 10:19
  SPEAKER_00: Their effect on citation selection is likely to depend on topic relevance, source quality, recency, accessibility, and the specific artificial intelligence system.
• 10:28
  SPEAKER_00: Google search documentation recommends using an author page, URL, and SAMAS links to help identify article authors.
• 10:35
  SPEAKER_00: For datasets, Google specifically recommends using an ORCID ID in the SAMAS property for a person creator.
• 10:42
  SPEAKER_00: A simple profile page could use structured data like this.
• 10:45
  SPEAKER_00: Structured data does not guarantee a search feature or artificial intelligence citation.
• 10:50
  SPEAKER_00: Google states that even correctly implemented structured data may not appear in search results.
• 10:54
  SPEAKER_00: Its value is that it gives machines a clear standard way to interpret the page.
• 10:59
  SPEAKER_00: Google Structured Data Guidance.
• 11:01
  SPEAKER_00: The practical conclusion is identifiers improve the path to evidence.
• 11:06
  SPEAKER_00: They do not replace evidence.
• 11:08
  SPEAKER_00: What current research says about artificial intelligence citations?
• 11:12
  SPEAKER_00: Research on artificial intelligence citation behavior has mainly studied citation accuracy, source selection, and verifiability.
• 11:20
  SPEAKER_00: It has not yet produced a reliable, general estimate for the effect of ORCID-linked identity strength.
• 11:26
  SPEAKER_00: A 2023 study of citations generated by ChatGPT found substantial rates of fabricated references and errors in real references.
• 11:35
  SPEAKER_00: The study examined 636 citations generated across 84 papers and found that citation problems remained even in the newer model tested.
• 11:45
  SPEAKER_00: Scientific reports study.
• 11:46
  SPEAKER_00: Another study of generative search engines found that many generated sentences were not fully supported by citations and that many citations did not adequately support the statements beside them.
• 11:58
  SPEAKER_00: Evaluating verifiability in generative search engines.
• 12:01
  SPEAKER_00: Recent research also shows that generative search systems differ in their source choices, source overlap, stability, and use of internal versus external knowledge.
• 12:11
  SPEAKER_00: Characterizing web search in the age of generative artificial intelligence.
• 12:15
  SPEAKER_00: These findings support a cautious position.
• 12:18
  SPEAKER_00: A DOI can reduce ambiguity about the cited work.
• 12:21
  SPEAKER_00: ORCID can reduce ambiguity about the person.
• 12:24
  SPEAKER_00: Affiliation identifiers can reduce ambiguity about the organization.
• 12:28
  SPEAKER_00: Contributor roles can improve attribution.
• 12:31
  SPEAKER_00: A public profile can provide context.
• 12:33
  SPEAKER_00: None of these signals guarantees that a system will retrieve or cite the work.
• 12:38
  SPEAKER_00: A research program to measure identity strength and artificial intelligence citation frequency.
• 12:43
  SPEAKER_00: Organizations should not claim that author hubs improve artificial intelligence citations until they measure the relationship directly.
• 12:50
  SPEAKER_00: Define an author identity strength score.
• 12:53
  SPEAKER_00: The following is a proposed score, not an official standard.
• 12:57
  SPEAKER_00: For the full table, please open this article on autopod.co.
• 13:01
  SPEAKER_00: Do not include citation counts, journal prestige, or institutional rank in this score.
• 13:06
  SPEAKER_00: Those factors should be treated as separate variables.
• 13:09
  SPEAKER_00: Including them would make it difficult to determine whether identity strength itself is related to citation frequency.
• 13:15
  SPEAKER_00: Measure more than one kind of citation.
• 13:18
  SPEAKER_00: A useful study should track at least six outcomes.
• 13:21
  SPEAKER_00: Work citation rate, the percentage of eligible answers that cite a specific work.
• 13:25
  SPEAKER_00: Author attribution rate, the percentage of answers that correctly name or attribute the work to the right author.
• 13:31
  SPEAKER_00: Valid identifier rate, the percentage of citations containing a DOI that resolves to the correct work.
• 13:37
  SPEAKER_00: Citation precision, the percentage of citations that actually support the statement made.
• 13:42
  SPEAKER_00: Citation recall, the percentage of important supporting sources that the system cites.
• 13:47
  SPEAKER_00: Cross-platform stability, the percentage of citations repeated across different systems or repeated runs.
• 13:54
  SPEAKER_00: A recent measurement framework for generative search separates citation selection from citation absorption, meaning whether a source merely appears in the citation list or actually contributes evidence to the answer.
• 14:05
  SPEAKER_00: This distinction should be included in future author hub studies, citation selection and citation absorption framework.
• 14:12
  SPEAKER_00: Study 1, observational correlation study.
• 14:15
  SPEAKER_00: This study asks, are works connected to stronger author identities cited more often by artificial intelligence systems after controlling for topic and publication factors?
• 14:24
  SPEAKER_00: Suggested design.
• 14:26
  SPEAKER_00: Sample 10,000 to 50,000 scholarly works.
• 14:29
  SPEAKER_00: Cover several disciplines, languages, and publication types.
• 14:33
  SPEAKER_00: Use Crossref, Datasite, OpenALECs, and public ORCID data.
• 14:38
  SPEAKER_00: Calculate an identity strength score for each author and work.
• 14:41
  SPEAKER_00: Create a set of questions for which the sampled works are relevant.
• 14:44
  SPEAKER_00: Run the questions across several citation-producing systems.
• 14:48
  SPEAKER_00: Repeat each question several times.
• 14:50
  SPEAKER_00: Record the date, system model setting, region, and cited sources.
• 14:54
  SPEAKER_00: Important control variables the study should control for.
• 14:57
  SPEAKER_00: Topic relevance, publication year, open access status, abstract availability, full text availability, venue, existing scholarly citation count, institutional reputation, language, page speed, and crawlability, work type, author name commonness, number of co-authors, query wording, search system.
• 15:16
  SPEAKER_00: The main statistical model could estimate a logistic regression or mixed effects model would be appropriate for a yes or no outcome, such as whether a work was cited.
• 15:24
  SPEAKER_00: A negative binomial model may be useful for citation counts because citation data are often unevenly distributed.
• 15:30
  SPEAKER_00: The result should be reported as an odds ratio with a confidence interval, not as a simple percentage.
• 15:35
  SPEAKER_00: For example, a one standard deviation increase in identity strength was associated with a change in citation probability after controlling for relevance, age, access, and prior scholarly attention.
• 15:47
  SPEAKER_00: The study should also report null results.
• 15:49
  SPEAKER_00: A finding of no relationship would be useful because it would show that identifiers improve data quality without necessarily changing artificial intelligence citation behavior.
• 15:58
  SPEAKER_00: Study 2.
• 15:59
  SPEAKER_00: Controlled metadata experiment.
• 16:01
  SPEAKER_00: An observational study may confuse identity strength with popularity.
• 16:05
  SPEAKER_00: A controlled experiment can isolate some effects.
• 16:08
  SPEAKER_00: Version A.
• 16:09
  SPEAKER_00: Public Web Experiment.
• 16:11
  SPEAKER_00: Create matched pages with the same substantive content, but different metadata treatments.
• 16:16
  SPEAKER_00: Basic author name only, author name plus institutional profile, author name plus ORCID link, author name ORCID DOI, affiliation identifier, and structured data.
• 16:28
  SPEAKER_00: Publish the pages at the same time and monitor search discovery, correct author matching, retrieval into artificial intelligence answers, citation frequency, citation correctness.
• 16:39
  SPEAKER_00: This experiment must be interpreted carefully.
• 16:42
  SPEAKER_00: Public systems may crawl pages at different times, and organizations cannot easily change crossref or data site records for an already published work.
• 16:51
  SPEAKER_00: Version Bontrolled Research Index.
• 16:54
  SPEAKER_00: Build a small retrieval system containing the same documents under different metadata conditions.
• 16:59
  SPEAKER_00: Very only author identifier, DOI, affiliation, contributor role, profile links, provenance fields.
• 17:07
  SPEAKER_00: Then ask the same questions and measure retrieval and citation selection.
• 17:11
  SPEAKER_00: This experiment gives better control, but it measures a research system rather than commercial platforms.
• 17:17
  SPEAKER_00: Study three Staggered Organizational Rollout.
• 17:21
  SPEAKER_00: The strongest practical design is a difference in differences study.
• 17:25
  SPEAKER_00: An organization can roll out author hubs and stages.
• 17:28
  SPEAKER_00: Department one receives the full program in October 2026.
• 17:32
  SPEAKER_00: Department two receives it in January 2027.
• 17:35
  SPEAKER_00: Department three receives it in April 2027.
• 17:38
  SPEAKER_00: All departments are measured before and after implementation.
• 17:42
  SPEAKER_00: Departments that have not yet received the program serve as temporary comparison groups.
• 17:46
  SPEAKER_00: Measure correct author attribution, DOI resolution, citation frequency, citation accuracy, search impressions, profile page visits, time needed to correct bad metadata.
• 17:58
  SPEAKER_00: This design is stronger than a simple before and after comparison because it helps separate the effect of the rollout from broader changes in artificial intelligence systems.
• 18:07
  SPEAKER_00: Hypotheses worth testing.
• 18:09
  SPEAKER_00: Organizations can pre-register these hypotheses.
• 18:12
  SPEAKER_00: H1.
• 18:12
  SPEAKER_00: Stronger identity links improve correct author attribution.
• 18:16
  SPEAKER_00: H2.
• 18:17
  SPEAKER_00: Complete DOI metadata improves valid citation rates.
• 18:20
  SPEAKER_00: H3.
• 18:21
  SPEAKER_00: Affiliation identifiers help most when authors have common names or move between institutions.
• 18:27
  SPEAKER_00: H4.
• 18:27
  SPEAKER_00: Scholar profiles improve discovery more than they improve final citation frequency.
• 18:32
  SPEAKER_00: H5.
• 18:33
  SPEAKER_00: Identity strength has a larger effect on long tail authors than on already famous authors.
• 18:38
  SPEAKER_00: H6.
• 18:39
  SPEAKER_00: The effect varies by system, topic, language, and publication type.
• 18:42
  SPEAKER_00: H7.
• 18:43
  SPEAKER_00: Freshness and topical relevance matter more than profile decoration.
• 18:48
  SPEAKER_00: What credentials should be published?
• 18:50
  SPEAKER_00: A profile should publish credentials that can be checked.
• 18:53
  SPEAKER_00: Each credential should have a credential type, the person who holds it, the issuing organization, a persistent identifier for the organization when available.
• 19:03
  SPEAKER_00: Start and end dates.
• 19:04
  SPEAKER_00: A verification source.
• 19:06
  SPEAKER_00: A status such as current, expired, disputed, or archived, a last reviewed date.
• 19:12
  SPEAKER_00: For the full table, please open this article on autopod.co.
• 19:16
  SPEAKER_00: Organizations should label the source of each credential.
• 19:19
  SPEAKER_00: Self-reported, institutionally verified, publisher verified, funder verified, imported from a registry, pending review.
• 19:28
  SPEAKER_00: Avoid publishing unsupported labels such as leading expert or world-renowned researcher.
• 19:33
  SPEAKER_00: Replace them with evidence, number and type of works, areas of research, verified roles, funded projects, editorial service, public data sensor software, institutional appointments.
• 19:45
  SPEAKER_00: This approach is more useful to both readers and machines.
• 19:48
  SPEAKER_00: Rollout plan for organizations.
• 19:51
  SPEAKER_00: Phase one.
• 19:52
  SPEAKER_00: Create the metadata policy.
• 19:53
  SPEAKER_00: Assign a small governance group with representatives from research administration, library services, information technology, communications, publishing or repository services, privacy, research integrity.
• 20:07
  SPEAKER_00: Create a short data policy that defines required fields, approved identifier systems, who may assert each field, which fields are public, how corrections are handled, how long updates should take, what happens when a person leaves.
• 20:20
  SPEAKER_00: Use a single data contract across systems.
• 20:23
  SPEAKER_00: At minimum, every person, work, organization, and credential record should have.
• 20:28
  SPEAKER_00: Phase 2.
• 20:29
  SPEAKER_00: Collect authenticated identifiers.
• 20:31
  SPEAKER_00: Ask researchers to connect their ORCID records through an authenticated process.
• 20:35
  SPEAKER_00: Do not make the institution responsible for guessing which ORCID ID belongs to a person.
• 20:40
  SPEAKER_00: At the same time, assign organization identifiers to departments and institutions.
• 20:45
  SPEAKER_00: Link staff records to ORCID.
• 20:47
  SPEAKER_00: Match existing publications to DOI records.
• 20:50
  SPEAKER_00: Flag name collisions.
• 20:51
  SPEAKER_00: Preserve alternate name forms.
• 20:53
  SPEAKER_00: Record consent and visibility settings.
• 20:55
  SPEAKER_00: Phase 3.
• 20:56
  SPEAKER_00: Improve DOI metadata.
• 20:58
  SPEAKER_00: For new works, require the publication workflow to collect full author names, authenticated ORCID IDs, affiliations, organization identifiers, contributor roles, funding identifiers, abstracts, references, licenses, version and update information.
• 21:15
  SPEAKER_00: For older works, begin with the most recent five years and the organization's highest use authors.
• 21:20
  SPEAKER_00: Crossref and data site should receive corrections and updates rather than only the original deposit.
• 21:25
  SPEAKER_00: DOI strings remain persistent while associated metadata can be maintained.
• 21:30
  SPEAKER_00: Crossref metadata maintenance.
• 21:32
  SPEAKER_00: Phase 4.
• 21:33
  SPEAKER_00: Build canonical author pages.
• 21:35
  SPEAKER_00: Each researcher should have one stable institutional page.
• 21:38
  SPEAKER_00: That page should use a permanent web address, show the current role, preserve previous roles with dates, link to ORCID, link to selected DOI records, link to public scholar profiles, show research areas, display a last updated date, use structured data, remain accessible to search crawlers, avoid requiring a login.
• 22:00
  SPEAKER_00: Recommends using clear author URLs and SAMAS links to help distinguish authors.
• 22:04
  SPEAKER_00: Google Author Markup Guidance.
• 22:06
  SPEAKER_00: Phase 5.
• 22:07
  SPEAKER_00: Federate and Monitor Profiles.
• 22:08
  SPEAKER_00: Link the institutional page to ORCID, Google Scholar Open Alex, Publisher Pages, Repository Records, Grant Pages, Public Professional Profiles where appropriate.
• 22:18
  SPEAKER_00: Use automated checks to find broken DOI links, missing ORCID links, duplicate authors, wrong co-author matches, old affiliations, expired credentials, missing update dates, mismatched names, retractions or corrections, governance for updates and role changes.
• 22:34
  SPEAKER_00: An author hub must preserve history.
• 22:36
  SPEAKER_00: It should not rewrite the past every time a person changes jobs.
• 22:39
  SPEAKER_00: Recommended source hierarchy.
• 22:41
  SPEAKER_00: For the full table, please open this article on autopond.co.
• 22:45
  SPEAKER_00: When a researcher changes institutions, do not delete the old affiliation from historical works.
• 22:50
  SPEAKER_00: Instead, add an end date to the former employment record, add the new affiliation with a start date, update the institutional profile.
• 22:58
  SPEAKER_00: Keep historical publication affiliations unchanged unless the publisher issues a correction.
• 23:03
  SPEAKER_00: Update the ORCID record through the proper source.
• 23:06
  SPEAKER_00: Redirect the old profile page to an archived or new page.
• 23:10
  SPEAKER_00: When a person changes name, use ORCID to connect name variants.
• 23:13
  SPEAKER_00: Add former or alternate names where the researcher agrees.
• 23:16
  SPEAKER_00: Do not change the author name in a published DOI record merely to make it match a current profile.
• 23:22
  SPEAKER_00: The purpose of the persistent identifier is to connect works across name changes without rewriting the publication record.
• 23:29
  SPEAKER_00: When a contributor's role changes, a contributor's role on a completed work should be treated as part of the historical record.
• 23:35
  SPEAKER_00: For example, a person's role on a published article should not change because they later become a department chair.
• 23:41
  SPEAKER_00: An editorial role should have a start and end date.
• 23:44
  SPEAKER_00: A grant role should be tied to the grant period.
• 23:47
  SPEAKER_00: A software role should be tied to the version or release.
• 23:50
  SPEAKER_00: If the original contribution record is wrong, use a formal correction process.
• 23:54
  SPEAKER_00: Do not silently override it.
• 23:56
  SPEAKER_00: When a credential expires, use clear status values.
• 23:59
  SPEAKER_00: Current, expired, renewed, suspended, disputed, archived.
• 24:03
  SPEAKER_00: Automated reminders should be sent before expiration.
• 24:06
  SPEAKER_00: Public profiles should show the status and date rather than leaving a stale credential visible without explanation.
• 24:12
  SPEAKER_00: When a researcher leaves the organization, the organization should preserve the historical profile, mark the former role clearly, retain links to published works, transfer profile ownership, set a redirect for the former page, remove private contact information, keep institutional assertions that were true during the former employment period, and when authorship is disputed, freeze the disputed field, preserve the original source, and record the dispute.
• 24:39
  SPEAKER_00: Do not remove an author or change contribution roles based only on an informal request.
• 24:44
  SPEAKER_00: The Research Integrity Office, Publisher, or Responsible Project Authority should determine the formal correction path.
• 24:52
  SPEAKER_00: Success measures Organizations should track operational quality before promising increased artificial intelligence visibility.
• 24:59
  SPEAKER_00: Suggested targets include 95% of active researchers have authenticated ORCID IDs.
• 25:05
  SPEAKER_00: 95% of new works include ORCID where available.
• 25:09
  SPEAKER_00: 95% of affiliations use a persistent organization identifier.
• 25:13
  SPEAKER_00: 100% of new DOI deposits pass metadata validation.
• 25:17
  SPEAKER_00: 100% of author pages contain a stable profile address.
• 25:21
  SPEAKER_00: 100% of role changes include dates.
• 25:24
  SPEAKER_00: Fewer than 1% of works are incorrectly assigned to an author.
• 25:28
  SPEAKER_00: Metadata corrections are completed within 30 days.
• 25:31
  SPEAKER_00: Current employment changes appear within 7 days.
• 25:34
  SPEAKER_00: Retractions and formal corrections are reflected promptly.
• 25:37
  SPEAKER_00: For the Artificial Intelligence Citation Study, track correct author attribution rate, valid DOI rate, citation precision, citation recall, work citation frequency, cross-system agreement, time from metadata correction to improved indexing, difference between profile discovery and final answer citation.
• 25:56
  SPEAKER_00: The organization should report these measures by discipline, language, career stage, and name commonness.
• 26:02
  SPEAKER_00: Otherwise, an average score may hide unequal outcomes.
• 26:05
  SPEAKER_00: Risks and safeguards.
• 26:06
  SPEAKER_00: Author hubs can create new risks if they become a gatekeeping system.
• 26:10
  SPEAKER_00: Organizations should not penalize researchers who keep some orca information private.
• 26:15
  SPEAKER_00: Treat a public Google Scholar profile as proof of quality.
• 26:18
  SPEAKER_00: Use citation counts as a substitute for peer review.
• 26:22
  SPEAKER_00: Publish personal contact details unnecessarily, infer expertise from institutional prestige alone.
• 26:27
  SPEAKER_00: Require every researcher to use the same public profile service.
• 26:31
  SPEAKER_00: Treat missing identifiers as evidence of misconduct.
• 26:34
  SPEAKER_00: Allow automated systems to overwrite records without review.
• 26:38
  SPEAKER_00: ORCID adoption and metadata quality vary by field, country, career stage, and publication type.
• 26:44
  SPEAKER_00: A fair system must support manual correction, privacy controls, name diversity, and people whose work is not represented well by journal publication metrics.
• 26:53
  SPEAKER_00: Conclusion.
• 26:54
  SPEAKER_00: A strong author hub is not a personal branding project.
• 26:57
  SPEAKER_00: It is a machine readable evidence system.
• 26:59
  SPEAKER_00: ORCID connects a person to a persistent identity.
• 27:02
  SPEAKER_00: Crossref and DataSight connect works to stable identifiers and rich metadata.
• 27:07
  SPEAKER_00: Research organization registry identifiers connect people and works to institutions.
• 27:12
  SPEAKER_00: Contributor roles show what each person actually did.
• 27:16
  SPEAKER_00: Institutional and scholar profiles provide public context.
• 27:20
  SPEAKER_00: Structured data helps search systems interpret the relationships.
• 27:24
  SPEAKER_00: Governance and version history show what is current, what was true in the past, and who confirmed each claim.
• 27:30
  SPEAKER_00: The likely benefit is not an automatic ranking boost.
• 27:33
  SPEAKER_00: The benefit is a stronger chain from person, work, organization, evidence.
• 27:38
  SPEAKER_00: Organizations should therefore make two commitments.
• 27:41
  SPEAKER_00: Standardize and maintain the metadata.
• 27:43
  SPEAKER_00: Measure artificial intelligence citation behavior rather than assuming an effect.
• 27:48
  SPEAKER_00: The most credible author hub is not the one with the most badges.
• 27:51
  SPEAKER_00: It is the one whose important claims are persistent, connected, sourced, current, and easy to verify.
• 27:57
  SPEAKER_00: All links to sources are available in the text version of this article.
• 28:01
  SPEAKER_00: You can find the full article at autopod.co.