The Sphinx build had dozens of RST parse errors and broken references preventing a clean docs rebuild, leaving the published HTML misaligned with current source.

RST / docfiles fixes

• conf.py — Added autodoc_mock_imports = ['gensim', 'rdflib'] and sphinx.ext.todo to eliminate import errors during autodoc
• reference/api_index.rst — Added :no-index: to all 98 automodule:: directives; these shadowed apidocs.rst and generated hundreds of duplicate-object warnings
• reference/algorithm_reference.rst — Removed 26 duplicate .. _label: anchors that collided with algorithm_roadmap.rst
• user_guide/recipes_and_workflows.rst — Renamed .. _recipes: → .. _user-recipes: (label collision)
• reference/layer_set_algebra.rst — Escaped bare * after = to prevent unclosed-emphasis parse error
• query_with_patterns.rst — Repaired 2 malformed grid tables and 24 broken section underlines
• query_zoo.rst, examples_reference.rst — Removed broken literalinclude blocks referencing non-existent files
• run_community_detection.rst, simulate_dynamics.rst — Replaced broken :download: refs with direct GitHub links
• multilayer_in_2min.rst — Fixed underline length mismatch

Python docstring fixes

Sphinx parses docstrings as RST; these patterns caused hard errors:

• |pipe| substitution references — |Z|, |Z_observed|, |z-score|, O(|V|) rewritten as plain prose across null_models.py, ast.py
• **kwargs bold markup — Wrapped as **kwargs in io/api.py
• '-|>' arrowstyle literal — Escaped to '-\\|>' in drawing_machinery.py
• Nested bullet lists in Napoleon Args: blocks — Flattened to prose in builder.py (aggregate, mutate), partition.py
• Continuation line indent — Fixed in result.py to_pandas docstring
• Glob pattern — *.edges → \*.edges in parsers.py

Rebuilt HTML

Docs rebuilt via sphinx-build -b html; output (zero errors) copied to docs/ for GitHub Pages. Remaining 132 warnings are all non-blocking: 65 toc.not_included notices, missing images in visualization guide, and external sklearn glossary references outside this codebase.