Computing Jaccard similarity between Wikidata items
df = aq.table({
item: ["Q1", "Q1", "Q1", "Q2", "Q2", "Q3", "Q3"],
prop: ["P31", "P27", "P106", "P31", "P69", "P21", "P31"]
})
df.view()
df1 = df.groupby("prop").pivot("item", { prop: (d) => op.valid(d.prop) })
df1.view()
df1
.array("Q1")
.reduce((sum, value, index) => sum + value * df1.array("Q2")[index], 0) /
df1
.array("Q1")
.reduce(
(sum, value, index) => sum + Math.max(value, df1.array("Q2")[index]),
0
)
cols = df1.select(aq.not("prop")).columnNames()
cols.map(
(d) =>
df1
.array(d)
.reduce((sum, value, index) => sum + value * df1.array("Q1")[index], 0) /
df1
.array(d)
.reduce(
(sum, value, index) => sum + Math.max(value, df1.array("Q1")[index]),
0
)
)
cols.reduce((result, d) => {
result[d] =
df1
.array(d)
.reduce((sum, value, index) => sum + value * df1.array("Q1")[index], 0) /
df1
.array(d)
.reduce(
(sum, value, index) => sum + Math.max(value, df1.array("Q1")[index]),
0
);
return result;
}, {})
jaccard = aq.from(
cols.map((e) =>
cols.reduce((result, d) => {
result[d] =
df1
.array(d)
.reduce((sum, value, index) => sum + value * df1.array(e)[index], 0) /
df1
.array(d)
.reduce(
(sum, value, index) => sum + Math.max(value, df1.array(e)[index]),
0
);
return result;
}, {})
)
)
jaccard.view()
query = `SELECT ?item ?itemLabel ?property ?propertyLabel
WITH {
SELECT ?item WHERE {
wd:Q7278 (^(wdt:P279*)/^wdt:P31) ?item.
?item wdt:P17 wd:Q142.
}
LIMIT 10
} AS %PARTIES
WHERE {
{
SELECT DISTINCT ?item ?property WHERE {
INCLUDE %PARTIES
?item ?p ?statement.
?statement ?ps ?ps_.
?property wikibase:claim ?p;
wikibase:statementProperty ?ps.
}
}
SERVICE wikibase:label { bd:serviceParam wikibase:language "fr,en". }
}`
output = fetch(
`https://query.wikidata.org/sparql?query=${encodeURIComponent(query)}`,
{ headers: { accept: "application/sparql-results+json" } }
).then((response) => response.json())
data = aq
.from(
output.results.bindings.map((d) => ({
item: d.item.value,
itemLabel: d.itemLabel.value,
property: d.property.value,
propertyLabel: d.propertyLabel.value
}))
)
.derive({
propertyId: (d) =>
op.replace(d.property, "http://www.wikidata.org/entity/", ""),
itemId: (d) => op.replace(d.item, "http://www.wikidata.org/entity/", "")
})
.derive({
property2: (d) => d.propertyLabel + " (" + d.propertyId + ")",
item2: (d) => d.itemLabel + " (" + d.itemId + ")"
})
.select("item2", "property2")
data.view()
data_qp = data
.groupby("property2")
.pivot("item2", { prop: (d) => op.valid(d.property2) })
data_qp.view()
qids = data_qp.select(aq.not("property2")).columnNames()
df_jaccard = aq
.from(
qids.map((e) =>
qids.reduce((result, d) => {
result[d] =
data_qp
.array(d)
.reduce(
(sum, value, index) => sum + value * data_qp.array(e)[index],
0
) /
data_qp
.array(d)
.reduce(
(sum, value, index) =>
sum + Math.max(value, data_qp.array(e)[index]),
0
);
return result;
}, {})
)
)
.assign({ qid: qids })
df_jaccard.view()
viewof qq = Inputs.select(qids)
df_jaccard.orderby(aq.desc(qq)).select("qid", qq).view()
import { aq, op } from "@uwdata/arquero"
One platform to build and deploy the best data apps
Experiment and prototype by building visualizations in live JavaScript notebooks. Collaborate with your team and decide which concepts to build out.
Use Observable Framework to build data apps locally. Use data loaders to build in any language or library, including Python, SQL, and R.
Seamlessly deploy to Observable. Test before you ship, use automatic deploy-on-commit, and ensure your projects are always up-to-date.