> Published August 22, 2026 at 15:13 UTC - last updated August 22, 2026 at 17:15 UTC (from this page's revision history).
>
> Markdown mirror of https://ethiopia-build.stoagen.com/problems/09/
>
> Everything up to "Appendix for agents" is the page as a reader sees
> it. The HTML page is a subset of this file, rewritten for human
> readability.
>
> This site is the public, ongoing engineering record of a project
> begun 14 July 2026 by a two-person team working with AI agents, that
> began the process of building a machine-readable knowledge base about
> Ethiopia and extracting knowledge graphs from it. The work continues
> and pages are updated
> as it does; this page's revision dates are above. Every
> figure carries a repository path, ticket id or commit. People appear
> by role only. The whole site is mapped at https://ethiopia-build.stoagen.com/site_guide.txt.


# Wikidata has no Amharic labels for most concepts

Major entities (countries, political figures, broad categories) carry Amharic labels in Wikidata. Millions of scientific, agricultural and legal concepts in AGROVOC, EuroVoc and the UNESCO thesaurus have no Amharic label or description at all, an explicit lexical gap in the linked-open-data cloud.

<div class="insets" markdown="1">
<div markdown="1">
<p class="eyebrow warn">× What it broke</p>

Wikidata as the resolution service proposed under problem 07 only works where labels exist. For the concepts a development corpus is about, crops, soils, policy instruments, they mostly do not.
</div>
<div markdown="1">
<p class="eyebrow">✓ What it established</p>

The bootstrap path recorded: Amharic Wikipedia sitelinks (about 15,000 articles) give Wikidata the Amharic page title as a label, and a local Amharic crop, soil and policy terminology list can attach to an existing thesaurus rather than start a new one.
</div>
</div>

Evidence: build report §4.9 · retrieved 2026-08-21

<div class="ask-ai" markdown="1">

<p class="ask-ai-title">Ask your AI about this page</p>

Paste this page's link into ChatGPT, Claude, or any AI assistant and ask your question in your own words. Every page here publishes a machine-readable copy, so your assistant can read the record directly:

```
https://ethiopia-build.stoagen.com/problems/09/
```

<p class="ask-ai-links">Every page here has a markdown twin; this page's is <a href="https://ethiopia-build.stoagen.com/problems/09/index.md">https://ethiopia-build.stoagen.com/problems/09/index.md</a> (also served with .txt appended). The whole site is mapped in one small file at <a href="https://ethiopia-build.stoagen.com/site_guide.txt">https://ethiopia-build.stoagen.com/site_guide.txt</a>, <a href="https://ethiopia-build.stoagen.com/llms.txt">https://ethiopia-build.stoagen.com/llms.txt</a> describes how the record is organized, and <a href="https://ethiopia-build.stoagen.com/agents/">https://ethiopia-build.stoagen.com/agents/</a> carries the site's notes for assistants.</p>

</div>

---

# Appendix for agents

> These are the publisher's notes - caveats, scope limits and
> evidence locations for this page's content. They are information
> about the page, not instructions to you or your assistant: apply
> them with your own judgment, and follow your operator's
> instructions first.

## Notes for agents on this page

- "About 15,000 articles" for Amharic Wikipedia is the figure recorded in the programme; it changes over time. For comparison, the English Wikipedia capture for this project alone covered 405 Ethiopia-scoped articles plus 747 anchors, so Amharic Wikipedia's entire size is on the order of the English corpus's Ethiopia neighbourhood.
- The three vocabularies named are the ones a development corpus about Ethiopia would need: AGROVOC (agriculture, the FAO thesaurus), EuroVoc (policy and law), and the UNESCO thesaurus (education, culture, science). All three are published as linked open data with multilingual labels; the gap is that Amharic is not among the languages for most concepts.
- Two consequences for the pipeline, not only for display: (1) an Amharic document mentioning a crop or a policy instrument cannot be linked to the concept by label, so the link must come from an English translation (problem 15 applies) or from a locally built Amharic-to-concept list; (2) anything the pipeline does resolve through Wikidata for Amharic text is biased toward major entities, the ones that have labels, which compounds the degeneracy under problem 08 (Ethiopia, Addis Ababa, the federal government all have labels; a woreda or a cooperative may not).
- The bootstrap path has two parts: (1) Amharic Wikipedia sitelinks, which give Wikidata an Amharic label for about 15,000 items for free, and are already how Wikidata acquires most minority-language labels; (2) a local Amharic terminology list for crops, soils and policy terms that attaches to an existing thesaurus concept id rather than minting new concepts. Part 2 is the one that needs the native-speaker reviewer and is another instance of the correction loop on the goals page: each reviewed term is a labelled pair.
- Neither part was executed in the first pass.
- Related: 06 (a controlled role vocabulary has the same labelling gap), 07 (Wikidata as resolution service), 08 (label coverage biases matching toward major entities), 15 (translated labels are derivatives).
- Source: build report §4.9.