Catalog exploration
Evidence mix
Count source types by label so a query can include papers, standards, posts, reports, and other source material deliberately.
Catalog source records include source_type. Use that field to ask whether a
label is grounded mostly in papers, standards, posts, reports, or a mix of
sources.
MIN_GUIDELINES=${MIN_GUIDELINES:-10}
MIN_SOURCES=${MIN_SOURCES:-5}
ORDER=${ORDER:-asc}Inspect source types
uvx chartcoach@latest catalog values references.source_type --format tableOutput:
| Source type | References |
|---|---|
article | 112 |
misc | 111 |
inproceedings | 38 |
incollection | 1 |
Rank labels by source mix
uvx chartcoach@latest catalog sql "
with label_sources as (
select
gl.label,
count(distinct gl.guideline_id) as guidelines,
count(distinct gs.reference_id) as sources,
count(distinct case
when gs.source_type in ('article', 'inproceedings', 'incollection')
then gs.reference_id
end) as published_sources,
count(distinct case
when gs.source_type = 'misc'
then gs.reference_id
end) as misc_sources
from guideline_labels gl
left join guideline_sources gs on gs.guideline_id = gl.guideline_id
group by gl.label
), scored as (
select
label,
guidelines,
sources,
published_sources,
misc_sources,
round(published_sources::double / nullif(sources, 0), 2) as published_share
from label_sources
where guidelines >= ${MIN_GUIDELINES}
and sources >= ${MIN_SOURCES}
)
select *
from scored
order by published_share ${ORDER}, guidelines desc, label
limit 5
" --format tableOutput with ORDER=asc:
| Label | Guidelines | Sources | Published sources | Misc sources | Published share |
|---|---|---|---|---|---|
basis:heuristic | 206 | 36 | 0 | 36 | 0.00 |
access:contrast:use | 10 | 14 | 1 | 13 | 0.07 |
access:keyboard:use | 10 | 18 | 2 | 16 | 0.11 |
quality:accessibility | 66 | 80 | 10 | 70 | 0.13 |
basis:accessibility | 50 | 80 | 10 | 70 | 0.13 |
Use ORDER=desc to list labels with the largest published-source share:
ORDER=descThe published bucket includes article, inproceedings, and incollection.
Keep misc in scope when the answer should include standards, practitioner
posts, reports, and other curated material.