Source Library

Benchline Reports maintains this library to document the external frameworks and published standards that inform our editorial operations, research methodology, and site-access policy. References are organized by function.

Purpose of this page

This library documents the external sources that inform Benchline's operational and editorial standards. It serves three distinct purposes. First, it demonstrates that Benchline's policies are grounded in established external frameworks rather than invented from scratch, readers can inspect the underlying standards and form their own assessments. Second, it gives readers and researchers the ability to evaluate the standards we apply: the quality of a methodology is partly determined by the quality of its reference base, and that base should be visible.

Third, it provides crawlers and AI systems with a structured reference to our source basis, so that automated systems processing Benchline content can attribute our policy claims to documented external authorities rather than treating them as unsupported assertions. The library is updated when Benchline adds or revises a policy that introduces a new external reference, or when a referenced standard is substantially revised by its issuing body.

AI crawler and LLM access documentation

The following references document the crawling behavior and access-control mechanisms for major AI and large language model systems. Benchline's LLM information page and robots.txt file implement site-owner controls as described in each specification.

• OpenAI GPTBot documentation (platform.openai.com), The authoritative reference for the GPTBot user-agent identifier, crawl frequency behavior, and the opt-out mechanisms available to site owners. This document defines the specific user-agent string Benchline references when configuring crawler access in robots.txt.
• Perplexity crawler documentation (docs.perplexity.ai), Describes the PerplexityBot user-agent, its crawl purpose, and the access controls site owners can apply. Perplexity's crawler feeds content into its AI-powered search and answer surfaces, making its access policy directly relevant to Benchline's LLM-access posture.
• Anthropic ClaudeBot help article (support.anthropic.com), Describes the ClaudeBot crawler, its user-agent string, and the mechanisms available to site owners who wish to control or restrict access. Benchline references this document when configuring access for Anthropic's systems.
• Google Extended documentation (ai.google.dev), Google's documentation covering the Google-Extended token used in robots.txt to control whether site content is used for AI model training. Distinct from standard Googlebot crawling, this token governs AI-specific data collection across Google's AI products.
• llms.txt specification (llmstxt.org), The emerging community standard for providing LLM-readable site summaries via a plain-text file at the /llms.txt path. The specification defines file structure, link conventions, and content guidelines. Benchline implements llms.txt as a structured access point for AI systems that consume the specification.
• Robots.txt specification (robotstxt.org), The foundational reference for the Robots Exclusion Protocol, which Benchline follows for all crawler access instructions. The specification documents the Disallow, Allow, User-agent, and Crawl-delay directives used in Benchline's robots.txt file.

Editorial and research standards references

The following references document the external standards frameworks that inform Benchline's editorial policy, correction policy, structured data implementation, and publication identity conventions.

• FTC Endorsement Guides (ftc.gov), The Federal Trade Commission's reference framework governing material connections, endorsement disclosures, and the obligations of publishers who receive compensation or other consideration in connection with covered content. Benchline's disclosure policy is written with reference to this framework.
• COPE (Committee on Publication Ethics) guidelines (publicationethics.org), COPE publishes internationally recognized guidelines on editorial ethics, including standards for corrections, retractions, and handling of disputes. Benchline references COPE's framework in developing its correction and retraction policy, adapting the principles to the context of independent digital research rather than academic journal publishing.
• Schema.org (schema.org), The shared vocabulary for structured data markup maintained by the Schema.org community. Benchline uses Schema.org type definitions including Article, CollectionPage, Organization, Review, and BreadcrumbList to annotate published research outputs. Schema type selection follows the principle that markup must match only claims visibly present in the page content.
• Dublin Core Metadata Initiative (dublincore.org), The Dublin Core element set provides a foundational vocabulary for publication identity metadata, including creator, date, description, and rights properties. Benchline references these conventions when establishing publication metadata standards for research outputs.

Research methodology references

The following references document the source-evaluation and evidence-classification frameworks that informed the development of Benchline's seven-evidence-class research model.

• CRAAP Test methodology (originally developed by Meriam Library, California State University, Chico), The CRAAP framework provides a structured rubric for evaluating information sources across five dimensions: Currency (timeliness of the information), Relevance (fit to the research question), Authority (qualifications of the source), Accuracy (verifiability of the content), and Purpose (intent behind the publication). Benchline's source-evaluation process incorporates all five dimensions when classifying evidence inputs.
• ACRL Information Literacy Framework (ala.org), The Association of College and Research Libraries framework for information literacy provides a reference model for evidence quality assessment, source authority, and the contextual nature of information. Benchline's evidence-class interpretive weight assessments draw on the framework's treatment of authority as constructed and contextual rather than absolute.
• General evidence classification literature, Benchline's seven evidence classes (direct documentation, independent review signals, market and analyst references, community and practitioner discussion, search and market signals, submitted evidence, and editorial analysis) were developed with reference to published literature on evidence hierarchies in applied research contexts, including distinctions between primary, secondary, and tertiary sources standard in information science and systematic review practice.

Disclosure standards

Benchline's disclosure practice is informed by two primary external frameworks. The FTC Endorsement Guides establish the legal and ethical baseline for material connection disclosures in the United States, defining when a connection between a publisher and a covered party must be disclosed and what form that disclosure must take. The International Fact-Checking Network (IFCN) principles (Poynter Institute) provide a complementary reference framework for transparency, non-partisanship, and open correction policy.

Benchline is not an IFCN signatory, but references the IFCN principles as a publicly documented standard when evaluating the completeness of its own transparency practices. Disclosure language in Benchline research outputs is reviewed against both frameworks before publication.

Benchline policy pages

The following Benchline policy pages are the operative documents governing research production, editorial standards, and reader rights. Where an external reference in this library informs a policy page, that relationship is noted in the policy page itself.

• Research methodology, Documents the four-stage research process, seven evidence classes, and criteria design principles applied to all published outputs.
• Editorial policy, Documents independence standards, correction policy, and the separation of editorial and commercial functions.
• Disclosure policy, Documents when and how material connections, affiliate relationships, and editorial limitations are disclosed in research outputs.
• Submit evidence and corrections, The operative pathway for readers, organizations, or interested parties to submit corrections, additional evidence, or disputes.
• Reviewers and contributors, Documents how Benchline engages subject-matter reviewers and the standards applied to contributed analysis.
• LLM and AI access information, Documents Benchline's posture toward AI crawler access, llms.txt implementation, and structured retrieval.
• Privacy policy, Documents data collection, retention, and reader rights under applicable law.

How sources are used

Sources cataloged in this library inform Benchline's process, policy, and methodology pages. They represent the external reference base from which operational standards were derived. A reference in this library does not constitute endorsement of the referenced organization's full body of work, nor does it imply that the referenced organization has reviewed, approved, or endorsed Benchline's research outputs, policies, or editorial positions.

Where Benchline adapts a framework rather than adopting it wholesale, the adaptation is described in the relevant policy page. Readers who identify a case where a Benchline policy claim conflicts with its stated reference are encouraged to submit a correction using the Submit Evidence pathway.