feat(ontology): add Flow H for Ontology & Schema Discovery and document Open Knowledge Graphs API

Author: aborruso

skills/ckan-mcp/SKILL.md

@@ -40,6 +40,8 @@ User asks about data
   +-- Asks about data quality? ------> Flow F (Quality)
   |
   +-- Wants best/most relevant? -----> Flow G (Relevance Ranking + Analysis)
+  |
+  +-- Wants to schema/annotate data? -> Flow H (Ontology & Schema Discovery)
 ```
 
 ## Flows
@@ -282,6 +284,48 @@ Example: "Compare these three traffic datasets"
 - `ckan_package_search` returns many loosely-matched results and you need to surface the closest ones
 - User wants a comparison or summary across multiple datasets
 
+### Flow H — Ontology & Schema Discovery
+
+Use when: the user wants to define a schema for a dataset, find existing standards
+for their domain, discover controlled vocabularies, or map dataset fields to
+semantic terms (DCAT, GeoSPARQL, Schema.org, SSN, Data Cube, etc.).
+
+This is relevant when the user:
+- asks "which ontology should I use for X?"
+- wants to make their data interoperable or linked-data ready
+- needs field names aligned with existing W3C/OGC/EU standards
+- asks "is there a vocabulary for X?"
+
+**Tool**: query the Open Knowledge Graphs API via `Bash` with curl.
+
+```bash
+# Search ontologies for a domain
+curl -s "https://api.openknowledgegraphs.com/ontologies?q=TOPIC&limit=5" | jq .
+
+# Narrow to a category (Government & Public Sector, Geospatial, Environment & Agriculture, ...)
+curl -s "https://api.openknowledgegraphs.com/ontologies?q=TOPIC&category=CATEGORY&limit=5" | jq .
+
+# Search across all types (ontologies + software)
+curl -s "https://api.openknowledgegraphs.com/search?q=TOPIC&limit=5" | jq .
+```
+
+See [references/open-knowledge-graphs.md](references/open-knowledge-graphs.md) for
+the full API reference and a complete end-to-end example (air quality sensor dataset
+→ SSN/SOSA ontology → field mapping).
+
+```
+Example: "I have a CSV with sensor readings — what schema should I use?"
+-> curl "https://api.openknowledgegraphs.com/ontologies?q=sensor+observation+measurement&limit=5"
+-> top result: SSN/SOSA (W3C) — score 0.69
+-> follow homepage: https://www.w3.org/TR/vocab-ssn/
+-> map CSV columns to sosa:Observation, sosa:Sensor, sosa:resultTime, sosa:hasResult
+
+Example: "Which vocabulary covers open government datasets?"
+-> curl "https://api.openknowledgegraphs.com/ontologies?q=open+data+government&limit=5"
+-> results: DCAT, NIEMOpen, Core Organization Ontology
+-> recommend DCAT (W3C) for dataset metadata, schema.org for web publishing
+```
+
 ## Key Rules
 
 ### Query Construction
@@ -419,3 +463,4 @@ curl -s -X POST "https://data.europa.eu/sparql" \
 - [`references/europa-api.md`](references/europa-api.md) — Read this for any query involving data.europa.eu: REST API patterns, country filtering, SPARQL examples, EU data themes and country codes.
 - [`references/tools.md`](references/tools.md) — Full `ckanapi` CLI equivalents for every MCP tool, with jq formatting patterns and DuckDB analysis examples. Read this when you need to replicate or extend tool behavior via Bash, or when the user needs to explore CSV resources directly.
 - [`references/hvd.md`](references/hvd.md) — High Value Datasets (EU Regulation 2023/138): API filters, the 6 thematic categories and sub-categories, country breakdowns, and HVD on national CKAN portals. Read this when the user asks about HVD or "dati ad alto valore".
+- [`references/open-knowledge-graphs.md`](references/open-knowledge-graphs.md) — Open Knowledge Graphs API: semantic search over 1,800+ ontologies, vocabularies, and taxonomies. Read this when the user wants to find existing schemas for a dataset, discover controlled vocabularies, adopt W3C/OGC standards (DCAT, SSN, GeoSPARQL...), or map dataset fields to semantic terms.

skills/ckan-mcp/references/open-knowledge-graphs.md

@@ -0,0 +1,141 @@
+# Open Knowledge Graphs API
+
+Semantic search over 1,800+ ontologies, vocabularies, taxonomies, and semantic
+software tools cataloged from Wikidata. Useful for finding existing standards
+and schemas when a user wants to model or annotate a dataset.
+
+**Base URL**: `https://api.openknowledgegraphs.com`
+
+## Endpoints
+
+| Endpoint | Description | Key params |
+|----------|-------------|------------|
+| `GET /` | API info + category list | — |
+| `GET /ontologies` | Search ontologies, vocabularies, taxonomies | `q` (required), `category`, `limit` |
+| `GET /software` | Search semantic software tools | `q` (required), `limit` |
+| `GET /search` | Search across all types | `q`, `category`, `type` (ontology\|software), `limit` |
+
+**Categories**: Life Sciences & Healthcare · Geospatial · Government & Public Sector ·
+International Development · Finance & Business · Library & Cultural Heritage ·
+Technology & Web · Environment & Agriculture · General / Cross-domain
+
+**Response fields** (per result): `score`, `title`, `wikidataId`, `description`,
+`types`, `category`, `homepage`, `licenses`, `partOf`, `latestVersion`, `releaseDate`
+
+## When to use
+
+Use this API when the user:
+- Has a dataset and wants to know which standard schema or ontology to adopt
+- Asks "is there a vocabulary for X?"
+- Wants to make their data more interoperable or linked-data ready
+- Needs field names / property URIs aligned with an existing standard
+
+## Complete example — air quality sensor dataset
+
+**Scenario**: user has a CSV with columns `station_id`, `timestamp`, `pm25`,
+`pm10`, `no2`, `temperature`, and wants to know which ontology to use.
+
+### Step 1 — search for relevant ontologies
+
+```bash
+curl -s "https://api.openknowledgegraphs.com/ontologies?q=sensor+observation+measurement&limit=5" | jq .
+```
+
+Result (abridged):
+
+```json
+{
+  "query": "sensor observation measurement",
+  "total": 5,
+  "results": [
+    {
+      "score": 0.6947755,
+      "title": "Semantic Sensor Network Ontology",
+      "wikidataId": "https://www.wikidata.org/wiki/Q62211950",
+      "description": "area of research",
+      "types": ["Ontology"],
+      "category": "Technology & Web",
+      "homepage": "https://www.w3.org/TR/vocab-ssn/"
+    },
+    {
+      "score": 0.6919018,
+      "title": "Extensible Observation Ontology",
+      "wikidataId": "https://www.wikidata.org/wiki/Q60675025",
+      "types": ["Ontology"],
+      "category": "General / Cross-domain",
+      "homepage": "https://github.com/NCEAS/oboe/"
+    }
+  ]
+}
+```
+
+**Top result**: [SSN — Semantic Sensor Network Ontology](https://www.w3.org/TR/vocab-ssn/)
+(W3C standard, highest score, Technology & Web).
+
+### Step 2 — refine with a domain-specific query
+
+```bash
+curl -s "https://api.openknowledgegraphs.com/ontologies?q=air+quality+environment&category=Environment+%26+Agriculture&limit=3" | jq .
+```
+
+```json
+{
+  "results": [
+    {
+      "title": "Environment Ontology",
+      "homepage": "http://environmentontology.org/",
+      "licenses": ["Creative Commons CC0 License"]
+    }
+  ]
+}
+```
+
+### Step 3 — follow the homepage to inspect the schema
+
+```bash
+# Fetch the W3C SSN spec page and extract key sections
+curl -s https://www.w3.org/TR/vocab-ssn/ | grep -oP '(?<=<h[23][^>]*>)[^<]+' | head -30
+```
+
+Or fetch and read the raw turtle/OWL namespace to extract classes and properties:
+
+```bash
+curl -sL https://www.w3.org/ns/ssn/ -H "Accept: text/turtle" | grep "^ssn:\|^sosa:\|rdfs:label" | head -40
+```
+
+Key SSN/SOSA classes and properties relevant to the air quality CSV:
+
+| CSV column | SSN/SOSA term | URI |
+|------------|---------------|-----|
+| `station_id` | `sosa:Sensor` | `http://www.w3.org/ns/sosa/Sensor` |
+| `timestamp` | `sosa:resultTime` | `http://www.w3.org/ns/sosa/resultTime` |
+| `pm25`, `pm10`, `no2` | `sosa:hasResult` → `sosa:Result` | `http://www.w3.org/ns/sosa/hasResult` |
+| `temperature` | `sosa:observedProperty` | `http://www.w3.org/ns/sosa/observedProperty` |
+| (observation row) | `sosa:Observation` | `http://www.w3.org/ns/sosa/Observation` |
+
+### Step 4 — check for software tools (optional)
+
+```bash
+curl -s "https://api.openknowledgegraphs.com/software?q=ontology+mapping&limit=3" | jq '.results[] | {title, homepage, latestVersion}'
+```
+
+### Complete recommendation to the user
+
+The **W3C SSN/SOSA ontology** (`https://www.w3.org/TR/vocab-ssn/`) is the right
+standard for sensor observation data:
+- Each row → `sosa:Observation`
+- `station_id` → `sosa:madeBySensor` (pointing to a `sosa:Sensor` instance)
+- `timestamp` → `sosa:resultTime` (ISO 8601 datetime)
+- `pm25` / `pm10` / `no2` → separate `sosa:Observation` instances each with `sosa:hasResult`
+- For the pollutants themselves, pair with **ENVO** (Environment Ontology, CC0) for
+  observed property URIs
+
+## Tips
+
+- `score` ranges 0–1: results above 0.7 are strong matches, 0.5–0.7 are relevant,
+  below 0.5 are loose.
+- If `homepage` is absent, use `wikidataId` to find the resource on Wikidata and
+  follow its "official website" or "described at URL" statements.
+- Use `category` filter to narrow to a domain and avoid off-topic ontologies.
+- Cross-check found ontologies against schema.gov.it (`mcp__schema-gov-it__*` tools)
+  if the dataset is for the Italian public administration context.