Strategic Product Report: Al.Qaeda.net Archival Reconstruction — "The Dossier Environment"

Full report

On this page

The following analysis details the strategic transition of the Al.Qaeda.net Archival Reconstruction from a highly interactive, read-only nostalgic workstation into a fully realized, local-first personal research environment. The proposed next major release—themed "The Dossier Environment"—aims to synthesize the strict epistemological boundaries of digital humanities archival standards with the robust, zero-dependency mechanics of modern browser-based storage architectures. By leveraging the "local-first" software paradigm, this evolution will ensure that the workstation not only preserves historical digital artifacts but provides a secure, private, and highly capable desktop for scholarly synthesis, annotation, and comparative analysis. The subsequent sections deconstruct the state of the art across digital archives and personal knowledge management systems, evaluating complex architectural trade-offs regarding local storage volatility, algorithmic text comparison, and W3C-compliant data provenance.

The State of the Art: Benchmark Analysis

To architect a workstation that provides a step-change in usefulness, it is critical to evaluate the current ecosystem of digital humanities platforms, personal knowledge management (PKM) systems, and retro-computing environments. The analysis reviews ten pivotal benchmarks to extract structural, functional, and aesthetic lessons.

Capability-Gap Analysis

The current baseline product succeeds as a highly immersive digital preservation exhibit. It features an immutable 425-node archive, exact hash-verified book sources, and flawlessly executed classic Windows interactions. However, a severe capability gap exists between "visiting" a static archive and "working" within a digital humanities laboratory. Currently, the user's interactions—such as bookmarks, research trails, and explicit history—are ephemeral or strictly bounded to passive navigation. To transition into a genuine research environment, the system lacks writable user-space, formal citation generation, deep-text parallel comparison, granular highlighting, and persistent, portable data export. Furthermore, while the current architecture persists some state locally, it lacks a unified schema for a user to extract their intellectual labor, physically move it to another machine, and seamlessly resume work without relying on third-party cloud synchronization.

Core Architectural Foundations

The Local-First Paradigm and Data Volatility

The directive mandates a strict local-only personal-data rule. Implementing "local-first" software ensures that the application responds instantly, works entirely offline, and grants the user ultimate ownership of their data15. The fundamental premise of local-first software is swapping the traditional client-server relationship: the local device is treated as the primary, authoritative copy of the data, rather than a mere cache16. However, relying exclusively on browser-based storage mechanisms, such as IndexedDB, introduces a critical architectural vulnerability: data volatility. Browser vendors, most notably Apple through WebKit and Safari, enforce stringent data eviction policies under the umbrella of Intelligent Tracking Prevention. Under WebKit's rules, all script-writable storage—including IndexedDB, the Cache API, and the Origin Private File System (OPFS)—is silently deleted if the user does not interact with the specific domain for seven days18. While Chromium-based browsers generally allow origins to utilize up to 60% of total disk space and only evict data under severe storage pressure, Safari's 7-day chronological eviction poses a fatal threat to an application positioned as a long-term scholarly research tool19. Mitigating this volatility requires a robust, two-pronged data ownership strategy. First, the application must utilize the File System Access API where supported, specifically leveraging navigator.storage.persist() to request explicit persistent storage permissions from the user. While this protects against pressure-driven eviction in Firefox and Chrome, it is frequently ignored by Safari's heuristic-based permission model21. Therefore, the second and more critical mitigation is standardizing a frictionless, high-speed physical export protocol. The user must be trained to treat their local operating system's file system—not the browser's internal IndexedDB—as the ultimate repository for their "Dossier."

The "Briefcase" Import/Export Protocol

To counteract browser eviction permanently, the software will frame "Saving" as a physical file export rather than a cloud synchronization event. Early Windows applications solved the problem of storing disparate data types (text streams, metadata, directories) within a single file by utilizing the Microsoft Compound File Binary Format (CFBF), acting essentially as a file system within a file23. The new release will philosophically mirror this retro concept, but utilize modern web standards to create a highly portable archive. When a user initiates an export, the system will execute a client-side packaging routine. Standard JavaScript ZIP libraries like jszip buffer the entire archive in the browser's memory, which is highly prone to crashing when handling large research datasets, and significantly blocks the main rendering thread25. The fflate library is the optimal choice for this architecture; it is exceptionally lightweight, supports asynchronous Web Worker execution, and streams data with a drastically reduced memory overhead25. The resulting export will contain standard JSON-LD files (Annotations, Settings, Binder drafts) packaged rapidly into a ZIP file—the "Briefcase." Upon returning to the workstation, the user selects their Briefcase, and fflate decompresses the payload locally, instantly repopulating the IndexedDB state without a single network request.

Source Integrity and the W3C Web Annotation Model

Maintaining epistemological source integrity demands a hard boundary between the immutable canonical archive and the mutable personal workspace. Modifying the original archival HTML to insert highlight spans would violate the principle of immutability. The W3C Web Annotation Data Model provides the ideal architectural separation. By treating annotations as directed graphs (serialized in JSON-LD) containing a Body (the user's note or highlight) and a Target (the immutable archive page), the system can overlay user data dynamically at runtime without ever altering the source document29. Crucially, the exact position of these annotations must survive varying screen sizes, text reflows, and potential minor markup corrections in future archival updates. Relying on strict DOM node offsets is highly brittle. The W3C TextQuoteSelector provides a resilient alternative by anchoring annotations using the exact quoted string, flanked by a contextual prefix and suffix4. When the workstation loads an annotated page, the anchoring algorithm searches for the exact string; if the DOM has shifted, it utilizes the prefix and suffix to perform fuzzy-matching, ensuring the user's highlights are reattached accurately4.

Parallel Comparison and Textual Diffing Algorithms

A cornerstone of scholarly analysis is the side-by-side comparison of texts, such as evaluating different translations or chronological revisions of a document. Implementing a document comparator requires selecting an appropriate text-diffing algorithm. The industry standard, the Myers diff algorithm, operates by finding the longest common subsequence (LCS) to generate the shortest possible edit script (the minimum number of insertions and deletions)33. While mathematically optimal, the Myers algorithm frequently produces unintuitive, scrambled results for human readers when paragraphs or code blocks are reordered, as it aggressively matches isolated common words or punctuation across distinct sections33. For a research environment, readability supersedes algorithmic minimalism. The "Patience Diff" algorithm—originally developed by Bram Cohen—resolves this by identifying lines that appear exactly once in both documents and using them as primary anchors, before recursively applying the diff to the spaces between them33. This prevents moved blocks of text from being interpreted as a chaotic mix of minor insertions and deletions, preserving the semantic boundaries of paragraphs. Implementing a line-based or word-level Patience Diff will provide researchers with a vastly superior tool for identifying nuanced historical discrepancies.

Archival Provenance and Metadata Standards

To elevate the workstation from a novelty into a rigorous academic tool, the metadata describing the archival nodes must adhere to established digital humanities standards. Exposing standard metadata allows researchers to trust the provenance of the digital artifacts. The implementation will utilize a triad of standards. First, Dublin Core provides a 15-element baseline (including Title, Creator, Subject, Date, and Identifier) for categorizing the fundamental properties of any resource37. Second, Encoded Archival Description (EAD), an XML standard maintained by the Society of American Archivists, will inform the hierarchical presentation of the finding aids and archive structures, moving beyond flat lists into a formal nested arrangement of series, sub-series, and items9. Finally, Encoded Archival Context for Corporate Bodies, Persons, and Families (EAC-CPF) will dictate how the system maps and displays the relationships between the individuals and organizations responsible for the creation of the records41. By surfacing this metadata in a dedicated "Record Inspector," the workstation firmly roots its aesthetic in rigorous archival science.

Comprehensive Capability Prioritization Matrix

To determine the most viable trajectory for the next major release, 25 distinct candidate capabilities were formulated based on the research mission. These candidates were scored against 15 strict criteria. Each criterion was graded on a scale of 1 to 3, where 3 represents the most favorable outcome (e.g., High Usefulness, High Local Feasibility, Low Implementation Complexity). Note: For Implementation Complexity, Maintenance Burden, and Failure Risk, a score of 3 indicates "Low/Favorable" and 1 indicates "High/Unfavorable".

Top 10 Recommended Features

Derived from the rigorous prioritization matrix, the following ten features will coalesce to forge the "Dossier Environment." They represent the optimal balance of profound researcher utility, uncompromising adherence to the local-only data mandate, and strict alignment with the retro-computing visual aesthetic.

1. Fflate-Powered Workspace ZIP Export/Import (The "Briefcase")

The cornerstone of the local-first strategy relies on physical data portability. Standard JavaScript archiving tools buffer entire payloads in memory, which is catastrophic when handling dense archival text and rich JSON-LD graph structures25. By integrating fflate, a highly optimized compression library that utilizes native CompressionStream APIs and Web Workers, the system can package the user's entire IndexedDB state—annotations, binder drafts, timelines, and UI configurations—into a ZIP file instantaneously, without freezing the main thread25. This "Briefcase" export empowers the researcher to physically move their work between offline computers, rendering Safari's 7-day data eviction threat entirely moot.

2. Local "Binder" Drafting Application

To graduate from an archive viewer to a scholarly laboratory, users require a dedicated staging ground. The Binder will manifest as a period-appropriate Multiple Document Interface (MDI) application, heavily styled after Windows 95/98 WordPad. Operating entirely within the browser's DOM, the Binder allows users to draft syntheses, drag-and-drop W3C annotations directly from the Archive Explorer, and compile evidence. Leveraging the optimistic UI pattern, every keystroke writes asynchronously to IndexedDB, ensuring zero data loss and instantaneous interface response times44.

3. W3C-Compliant Annotation Engine

A profound leap in utility will come from the ability to mark up the immutable archive. Adhering to the W3C Web Annotation Data Model, these annotations will exist as standalone JSON-LD entities within the user's local database, projecting visually onto the archive via a rendering layer29. To anchor these highlights, the system will utilize the TextQuoteSelector standard. By storing the exact string of the user's highlight alongside a 32-character prefix and suffix, the application can algorithmically re-attach highlights even if the underlying HTML elements are restructured during future archive updates4.

4. Side-by-Side Patience Diff Comparator

Evaluating contradictions across documents—such as comparing early drafts of a manifesto to a finalized translation—is a critical DH task. The workstation will introduce a split-pane Document Comparator. Eschewing the standard Myers diff algorithm, which minimizes the edit script at the cost of human readability, the comparator will deploy a "Patience Diff" algorithm33. By matching unique anchor lines first, Patience Diff keeps reordered paragraphs intact and highlights semantically meaningful changes, vastly accelerating the researcher's ability to parse complex textual divergences35.

5. CSL-JSON Citation Desk

Academic integrity relies on rigorous sourcing. The Citation Desk will function as an internal reference manager, generating correctly formatted citations (Chicago, APA, MLA) for any node within the archive. By utilizing the Citation Style Language JSON (CSL-JSON) schema—a standardized data model mapping item types (e.g., "book", "article") and variables to citation processors—the workstation can instantly compile a bibliography of the user's research trail3. These bibliographies can be seamlessly integrated into the Binder or exported as standalone HTML packets.

6. WAI-ARIA Compliant Treeview and Keyboard Navigation

Accessibility is paramount, even within a nostalgic interface. The core Archive Explorer relies on deeply nested hierarchical folders. Upgrading this component to comply with the WAI-ARIA treeview specification ensures that screen readers accurately announce node depth and expansion states46. Furthermore, implementing strict keyboard focus management (using Up/Down arrows to traverse visible nodes, and Right/Left to expand/collapse) will provide power users and disabled researchers with an efficient, mouse-free navigation paradigm that mirrors native desktop operating systems46.

7. IndexedDB State Manager with Schema Versioning

Local-first software introduces the complex burden of managing client-side database migrations. As the workstation evolves in future releases, the shape of the stored JSON objects will inevitably change. Implementing a rigorous IndexedDB state manager ensures that when a returning user imports a two-year-old "Briefcase" ZIP, the system detects the legacy schema and gracefully executes sequential migration scripts (onupgradeneeded)49. This defensive architecture guarantees that the user's intellectual labor is never corrupted by software updates.

8. CHM-Style Retro Help Viewer

Modern, intrusive onboarding modals fundamentally shatter the immersion of a retro environment. Instead, discoverability and documentation will be delivered through a dedicated Help Viewer meticulously designed to emulate the classic Microsoft HTML Help executable (hh.exe)7. The CHM viewer layout—featuring a resizable split-pane with a hierarchical table of contents on the left and HTML rendering on the right—remains one of the most cognitively efficient patterns for navigating dense technical documentation7.

9. Dublin Core & EAD Meta-Inspector

To establish trust in the archival materials, the underlying metadata must be visible and standardized. The Meta-Inspector application will expose the metadata of the currently active node using the 15-element Dublin Core standard (Title, Creator, Subject, Date, Format, etc.)37. Furthermore, the provenance and hierarchical context of the documents will be articulated using Encoded Archival Description (EAD) parameters9, and authorial relationships will be mapped utilizing the Encoded Archival Context (EAC-CPF) schema41. This grounds the virtual environment in authentic digital humanities practices.

10. Z-Index Stacking Context Manager

To support the intense multitasking of a true research environment, users must be able to open multiple Archive pages, the Binder, and the Citation Desk simultaneously. Operating a Multiple Document Interface (MDI) inside a single browser DOM requires a sophisticated z-index stacking context registry6. The window manager will be refactored to dynamically elevate the z-index of focused windows, handle modal overlays appropriately, and manage keyboard focus trapping, ensuring that the retro OS behaves identically to a native 32-bit operating system without visual artifacting or focus loss6.

Features Explicitly NOT to Build

To preserve the architectural integrity and mission parameters of the workstation, several modern paradigms must be actively avoided:

1. CRDT-Based Real-Time Collaboration: While Conflict-Free Replicated Data Types (CRDTs) represent the vanguard of collaborative local-first software, they carry immense computational overhead regarding vector clocks and tombstone management15. This workstation is designed as a solitary, private scholarly environment. Implementing multi-user syncing mechanisms introduces unnecessary complexity and network dependency, directly contradicting the isolated dossier objective.

2. Cloud Synchronization & Telemetry: Absolutely forbidden by the local-only personal-data rule. No user data, analytics, behavioral tracking, or telemetry will ever leave the browser boundary.

3. Modern UI Overlays or "Material Design": Immersive integrity dictates that all new functional layers (even complex features like text diffing or metadata inspection) must strictly conform to the 90s-era aesthetic. Drop shadows, border-radiuses, and flat design paradigms are prohibited in favor of 3D grey bevels, MS Sans Serif typography, and distinct window chrome.

4. Full-Text NLP / Machine Learning Indexing: Attempting to execute heavy Natural Language Processing (NLP) models or LLM-based vector embeddings locally via WebAssembly would severely degrade performance on lower-end devices and bloat the application footprint. The workstation will rely on highly optimized, deterministic RegEx search and exact-match indexing to maintain instantaneous responsiveness.

User Journeys within the Dossier Environment

The Casual Reader: A user visits the site out of historical curiosity. They navigate the Archive Explorer, reading various manifestos and analytical nodes. They utilize the CHM-style Help Viewer to familiarize themselves with the interface mechanics. They bookmark several pages and close the browser. If they return within the week, their bookmarks remain seamlessly loaded via IndexedDB. The interaction is frictionless, requiring zero account creation or onboarding overhead. The Serious Researcher: An academic opens the workstation to conduct a comparative analysis of primary sources. They utilize the Patience Diff tool to load two distinct translations of an archival text, rapidly identifying semantic divergences36. They highlight these contradictions using the W3C Web Annotation engine, tagging them with custom metadata. Opening the local Binder application, they drag and drop these annotations into a draft, utilizing the Citation Desk to append flawlessly formatted CSL-JSON Chicago-style references55. At the conclusion of their session, they initiate an export. The workstation flashes an era-accurate progress bar as fflate packages their workspace, delivering a ZIP file to their local machine for immutable safekeeping. The Returning Researcher: Weeks later, the researcher returns to the URL. Because their browser’s storage was cleared by aggressive OS maintenance or Safari's 7-day eviction policy, the workstation loads in a pristine, empty state. Undeterred, they click "Import Briefcase" and upload their previously saved ZIP file. The system operates entirely offline; fflate decompresses the archive locally27. The IndexedDB Schema Manager instantly validates the imported JSON-LD structures against the current software version, upgrading any legacy data arrays50. Within seconds, their Binder drafts, highlighting layers, and custom taxonomies are repopulated, allowing research to continue without interruption. User Moving Work Between Computers: A user begins synthesizing a dossier on a secure desktop workstation in a university library. They export their Briefcase ZIP, transfer it to an encrypted USB drive, and physically carry it to their personal laptop. Opening the web application on the laptop, they import the ZIP. Because the application logic relies on standardized JSON schemas and strict offline-first principles, the transition is flawless. No central server orchestration, authentication handshake, or cloud synchronization is required. The data remains entirely within the user's physical custody.

Period-Appropriate UI and Accessibility Strategy

The visual language of the application must remain rigidly locked to the classic Windows 95/98 desktop UI paradigm. All newly introduced research tools will be mapped to familiar historical constructs. The "Patience Diff Comparator" will utilize a standard grey split-pane window with classic scrollbars. The "Citation Desk" will be modeled after a complex Windows property sheet featuring multi-tabbed dialogues and inset group boxes. Simultaneously, adhering to modern accessibility standards within this retro aesthetic is non-negotiable. The WAI-ARIA treeview implementation for the Archive Explorer ensures that screen readers can accurately interpret the complex hierarchical depth of the finding aids46. Furthermore, an era-appropriate "Display Properties" dialog will be introduced, allowing visually impaired users to toggle a high-contrast mode and adjust text scaling without shattering the illusion of the retro operating system.

Risks and Architectural Mitigations

1. Data Eviction (Safari/WebKit): As detailed, the silent background deletion of IndexedDB by WebKit is the most severe threat to a local-first application18.

• Mitigation: The user interface must feature a prominent, period-accurate "System Tray" indicator warning the user of unsaved volatile state. The application will proactively prompt the user to generate a Briefcase ZIP export at the end of lengthy sessions, relying on education and physical export rather than fighting browser heuristics.

2. Schema Migration Failures: As the application evolves, updating the internal logic could inadvertently corrupt user data when they import an older Briefcase ZIP.

• Mitigation: Implement a rigorous migration pipeline mimicking the architecture of Dexie.js, ensuring flawless backward compatibility. The system will maintain a strict ledger of schema versions, sequentially applying onupgradeneeded transformations to legacy JSON objects before committing them to the active IndexedDB store50.

3. Memory Leaks in the MDI Environment: Opening dozens of virtual windows (Archive pages, Binders, Diff tools) inside a single browser tab can cause severe DOM bloat and garbage collection lag.

• Mitigation: Implement rigorous React/Vanilla DOM unmounting protocols when virtual windows are closed. For extensive text documents, implement virtual rendering (windowing) to ensure that only the text nodes currently visible within the scrolling viewport are painted to the DOM.

Sequencing and Implementation Phases

Phase 1: Foundation (Data & Storage Architecture)

• Integrate the fflate compression library and build the Briefcase ZIP export/import pipeline.
• Establish the IndexedDB schema architecture and the sequential local versioning manager.
• Implement OPFS integration to request persistent storage quotas from the browser.

Phase 2: Primary Features (The Workspace)

• Develop the local Binder application for rich text drafting and collation.
• Implement the W3C Web Annotation engine, specifically tuning the TextQuoteSelector fuzzy-matching algorithms.
• Refactor the Window Manager to support robust z-index stacking contexts and MDI focus trapping.

Phase 3: Refinement (Scholarly Research Tools)

• Build the Patience Diff Document Comparator for side-by-side textual analysis.
• Integrate the CSL-JSON Citation Desk and bibliography generator.
• Develop the CHM-style Retro Help Viewer to onboard users to the new capabilities.

Phase 4: Experimental & Polish (Metadata & Access)

• Implement the Dublin Core, EAC-CPF, and EAD metadata inspectors.
• Finalize WAI-ARIA keyboard navigation patterns across all hierarchical UI components.
• Deploy the Sandboxed Local File Opener, rigorously testing iframe security boundaries.

Objective Acceptance Criteria

To deem the "Next Major Release" ready for production deployment, the following metrics must be objectively met during quality assurance:

1. Total Offline Independence: The application must achieve a 100% functional load sequence (including full-text searching, annotation rendering, and diff generation) with the browser's network tab set to "Offline," relying exclusively on the Service Worker cache.

2. Export/Import Roundtrip Fidelity: A simulated 50-megabyte user workspace (containing 2,000 W3C annotations, 100 Binder drafts, and dense custom taxonomies) must compile into a ZIP export in under 3.5 seconds using fflate on standard hardware. Upon re-import, the payload must pass a cryptographic hash check ensuring 0% data corruption or schema deviation.

3. UI Rendering Performance: Opening 12 simultaneous MDI windows (including complex graph renderings and split-pane text comparators) must not drop the browser's rendering frame rate below 60fps, ensuring the retro OS feels as responsive as a native application.

4. Absolute Immutability Verification: Automated integration tests must systematically verify that no combination of user actions—whether within the Binder, Annotation Engine, or Diff Comparator—can alter the byte-level hash of the canonical archival source HTML stored in the core application bundle.

Works cited

1. The Ultimate Guide to Local-First Software \- Dench, https://www.dench.com/blog/ultimate-guide-local-first-software

2. citeproc-js bug? \- Zotero Forums, https://forums.zotero.org/discussion/62288/citeproc-js-bug

3. CSL-JSON as an Interchange Format \- GROBID Tools, https://grobid.org/formats/csl-json

4. Annotation Systems for Evolving Documents \- wal.sh, https://wal.sh/research/annotation-systems/

5. DustinBrett/awesome-OS \- GitHub, https://github.com/DustinBrett/awesome-OS

6. Create a window manager with React \- DEV Community, https://dev.to/jbdemonte/create-a-window-manager-with-react-3mak

7. About the HTML Help Executable Program | Microsoft Learn, https://learn.microsoft.com/en-us/previous-versions/windows/desktop/htmlhelp/about-the-html-help-executable-program

8. Microsoft Compiled HTML Help \- Wikipedia, https://en.wikipedia.org/wiki/Microsoft\_Compiled\_HTML\_Help

9. Encoded Archival Description (EAD) \- Metadata Standards Index, https://msi.dublincore.org/standards/ead

10. Other Web sites related to Encoded Archival Description, https://www.loc.gov/ead/eadother.html

11. dh-tech/awesome-digital-humanities \- GitHub, https://github.com/dh-tech/awesome-digital-humanities

12. DH Tools | DIGITAL HUMANITIES, https://digitalhumanities.duke.edu/doing-dh/dh-tools

13. The Why and How of Middleware \- DHQ Static, https://dhq-static.digitalhumanities.org/pdf/000248.pdf

14. Annotate \- Omeka S, https://omeka.org/s/modules/Annotate/

15. Local-First Software: Principles, Patterns, and Technologies \- wal.sh, https://wal.sh/research/local-first

16. Local-first software: You own your data, in spite of the cloud, https://www.inkandswitch.com/essay/local-first/

17. Local-first Software \- Ink & Switch, https://www.inkandswitch.com/local-first-software/

18. Updates to Storage Policy \- WebKit, https://webkit.org/blog/14403/updates-to-storage-policy/

19. How Much Chat History Can Your Browser Store? IndexedDB & OPFS, https://secret-chat.ai/blog/browser-storage-limits-indexeddb-opfs/

20. Storage for the web | Articles, https://web.dev/articles/storage-for-the-web

21. Storage quotas and eviction criteria \- Web APIs \- MDN Web Docs, https://developer.mozilla.org/en-US/docs/Web/API/Storage\_API/Storage\_quotas\_and\_eviction\_criteria

22. Origin Private FileSystem Question that is both on and off topic. Is, https://news.ycombinator.com/item?id=39222328

23. Compound File Binary Format \- Wikipedia, https://en.wikipedia.org/wiki/Compound\_File\_Binary\_Format

24. \[MS-CFB\]: Introduction \- Microsoft Learn, https://learn.microsoft.com/en-us/openspecs/windows\_protocols/ms-cfb/50708a61-81d9-49c8-ab9c-43c98a795242

25. zip.js/BENCHMARKS.md at master · gildas-lormeau/zip.js · GitHub, https://github.com/gildas-lormeau/zip.js/blob/master/BENCHMARKS.md

26. Limitations of JSZip, https://stuk.github.io/jszip/documentation/limitations.html

27. fflate \- NPM, https://www.npmjs.com/package/fflate

28. Architectural Strategies for ZIP Compression in Modern Web, https://npm-compare.com/pako,jszip,fflate,adm-zip,@zip.js/zip.js,zip-lib

29. Web Annotation Data Model \- W3C on GitHub, https://w3c.github.io/web-annotation/model/wd/

30. Web Annotation Protocol \- W3C, https://www.w3.org/TR/annotation-protocol/

31. Selectors and States \- W3C, https://www.w3.org/TR/selectors-states/

32. Web Annotation Vocabulary \- W3C, https://www.w3.org/TR/annotation-vocab/

33. What diff algorithms are available and which should I use?, https://inventivehq.com/blog/what-diff-algorithms-are-available-and-which-should-i-use

34. How Diff Algorithms Actually Work (And Why Yours Might Be Wrong), https://dev.to/alphashark/how-diff-algorithms-actually-work-and-why-yours-might-be-wrong-3c74

35. When to Use Each of the Git Diff Algorithms \- Technical Scratchpad, https://luppeng.wordpress.com/2020/10/10/when-to-use-each-of-the-git-diff-algorithms/

36. The patience diff algorithm \- The If Works \- James Coglan, https://blog.jcoglan.com/2017/09/19/the-patience-diff-algorithm/

37. Description and Cataloging of XML-Data Schemas \- DCMI, https://www.dublincore.org/specifications/dublin-core/dc-xml-data-description/

38. DCMI: Metadata Basics \- Dublin Core Metadata Initiative, https://www.dublincore.org/resources/metadata-basics/

39. Dublin Core Metadata Record: A Fully Worked Example \- CASRAI, https://casrai.org/guides/dublin-core-metadata-record-example

40. Encoded Archival Description: An Introduction and Overview, https://www.dlib.org/dlib/november99/11pitti.html

41. Encoded Archival Context \- Wikipedia, https://en.wikipedia.org/wiki/Encoded\_Archival\_Context

42. Corporate Bodies, Persons, and Families (EAC-CPF), https://standards.openpreservation.org/eac

43. Corporate bodies, Persons, and Families (EAC-CPF), https://www2.archivists.org/node/23669

44. Offline-first frontend apps in 2025: IndexedDB and SQLite in the, https://blog.logrocket.com/offline-first-frontend-apps-2025-indexeddb-sqlite/

45. Citation Style Language JSON Schema \- FAIRsharing, https://fairsharing.org/10.25504/FAIRsharing.pwk71v/

46. Tree View Pattern | APG | WAI \- W3C, https://www.w3.org/WAI/ARIA/apg/patterns/treeview/

47. Keyboard Navigation Patterns for Complex Widgets \- UXPin, https://www.uxpin.com/studio/blog/keyboard-navigation-patterns-complex-widgets/

48. Developing a Keyboard Interface | APG | WAI \- W3C, https://www.w3.org/WAI/ARIA/apg/practices/keyboard-interface/

49. IndexedDB: A Comprehensive Overview for Frontend Developers, https://medium.com/@shashika.silva88/indexeddb-a-comprehensive-overview-for-frontend-developers-6b47a9f32e23

50. Understanding the basics \- Offline-First Database | Dexie.js \- Dexie.js, https://dexie.org/docs/Tutorial/Understanding-the-basics

51. Make your CHM Help Files show HTML5 and CSS3 content, https://weblog.west-wind.com/posts/2012/feb/15/make-your-chm-help-files-show-html5-and-css3-content

52. Stacking context \- CSS \- MDN Web Docs, https://developer.mozilla.org/en-US/docs/Web/CSS/Guides/Positioned\_layout/Stacking\_context

53. Z-index and stacking contexts \- web.dev, https://web.dev/learn/css/z-index

54. Consistent Local-First Software: Enforcing Safety and Invariants for, https://www.researchgate.net/publication/384820390\_Consistent\_Local-First\_Software\_Enforcing\_Safety\_and\_Invariants\_for\_Local-First\_Applications

55. CSL-JSON — citeproc-js 1.1.73 documentation \- Read the Docs, https://citeproc-js.readthedocs.io/en/latest/csl-json/markup.html

56. IndexedDB and Web Workers: A Guide to Offline-First Web Apps, https://blog.adyog.com/indexeddb-and-web-workers-a-guide-to-offline-first-web-apps/