OBI means Ontological Bayesian Intelligence. It is an ontological and Bayesian reasoning infrastructure for uncertainty-aware robotic and humanitarian systems.
The repository contains a static documentation toolchain that converts plain text, Markdown, LaTeX, and BibTeX source material into consistent HTML pages, a client-side search index, bibliography outputs, and static publishing artifacts.
OBI should not be described as conscious AI. It is not a claim of machine consciousness, sentience, or personhood. The project language should stay focused on reasoning infrastructure, uncertainty modeling, ontologies, Bayesian inference, semantic graphs, robotics interfaces, and humanitarian real-world assistance systems.
.
├── .circleci/
├── .github/workflows/
├── docs/
│ ├── archive/
│ ├── public/
│ └── source/
├── nw/
├── obi/
├── scripts/
├── package.json
├── pyproject.toml
└── README.md
Source documents live under docs/source. Generated static output is written to docs/public. Research ZIP bundles are preserved under docs/archive.
Node.js is used for the documentation build. Python remains available for future research tooling and the package is still installable.
npm install
pip install -e .The Node build currently has no external runtime dependencies, so it is easy to run from PowerShell and CI.
npm run clean
npm run docs:populate
npm run docs:build
npm run docs:validate
npm run docs:serve
npm run docs:serve:html
npm run publish:checknpm run docs:populate extracts ZIP bundles from docs/archive, converts extracted text/LaTeX/PDF material into Markdown under docs/source/md, and writes standalone source HTML under docs/source/html.
npm run docs:build generates:
docs/public/index.htmldocs/public/papers.htmldocs/public/bibliography.htmldocs/public/search.jsondocs/public/sitemap.xmldocs/public/robots.txtdocs/public/manifest.jsondocs/public/assets/obi.cssdocs/public/assets/obi.js- copied PDFs under
docs/public/pdf - copied source HTML under
docs/public/source-html
Place source files in these folders:
.texindocs/source/tex.bibindocs/source/bib.mdindocs/source/md.pdfindocs/source/pdf.txtindocs/source/txt- images and diagrams in
docs/source/assets - generated archive/PDF HTML in
docs/source/html
The build converts source pages into static HTML under docs/public/sources.
The population step uses Python and prefers Pandoc for LaTeX-to-Markdown and Markdown-to-HTML conversion. The script also uses Python Markdown when it is installed, and uses pdfplumber or pypdf for PDF text extraction.
MiKTeX or another LaTeX distribution is optional for future PDF compilation. The HTML path does not require a TeX engine, but the generated docs/source/html/manifest.json records whether Pandoc, pdflatex, xelatex, lualatex, and bibtex were available.
Math syntax is preserved and rendered in HTML with MathJax. Image Markdown is rendered with alt text and responsive sizing so certificate images, diagrams, and extracted figures remain accessible in the HTML view.
To inspect the raw generated HTML directly:
npm run docs:serve:htmlOpen http://127.0.0.1:8000/html/.
BibTeX files belong in docs/source/bib. The build generates docs/public/bibliography.html and docs/public/bibliography.json.
LaTeX citations are linked when the citation key exists in the loaded BibTeX entries. Missing keys are shown as unresolved citations but do not fail the build.
The desktop wrapper lives in nw/package.json and points at ../docs/public/index.html.
npm run docs:build
npm run nw:startThe user-facing phrase may be node-webkit, but the implementation uses modern NW.js naming.
CircleCI installs Node.js dependencies, cleans generated docs, builds the static site, validates required outputs, and stores docs/public as an artifact.
The GitHub Actions workflow in .github/workflows/publish.yml runs on pushes to main, builds the documentation, validates it, uploads docs/public as a Pages artifact, and deploys it to GitHub Pages.