Table-of-contents extraction
List the titles, levels and locations of a text’s sections.
Definition
Recovering the outline of a work: every section heading, how deeply it is nested, and where it starts. In Tibetan commentaries the outline (sa bcad) is usually embedded in the running text rather than set as headings, so the task includes recognizing it.
Input and output
- Input: The full text of a work.
- Output: A list of entries in reading order, each with its title as written, its level, and its start position or page.
Examples
None yet. Add a short real example with its correct answer, and say where it comes from.
Edge cases
- Outlines announced in the text (“this has two parts”) whose parts appear pages later.
- Headings that look like ordinary sentences.
- Texts with both chapter titles and a sa bcad outline.
Languages and traditions
- Tibetan (bo): measured; see the benchmark below.
How it is measured
Benchmark: toc-extraction, Tibetan, version 2.0.0, scored by structure F1. Leaderboard. Best skill: toc-extraction-tibetan.
Help wanted
Benchmarks in other languages, and more Tibetan genres (sutras, collected works, histories). A first benchmark can be small: twenty to fifty items. See how to submit a benchmark.
Related tasks
Found in workflows
- Preparing a digital edition from a template repo (Sitar)
- Translating a Buddhist text (Kurt Keutzer)

BuddhistAI Knowledge base