The Carpathian data graph
Ask the AI guide «Where can I get water near Hoverla?» and then «Where can I sleep there?» — it understands that «there» still means Hoverla. The search finds «Goverla» spelled the wrong way. And a trail page shows which springs and huts you will pass. Behind all of this is a data graph: a map of the links between everything on Karpaty.Rocks.
The graph today
How many places and links between them the site knows right now. The numbers change every time a new page appears.
- 29,287nodes — places, trails, dishes, species
- 275,558links between them
- 59,902names and spellings
- 3,695peaks
- 9,926springs
- 3,928huts and shelters
- 2,707trails
- 2,256villages and towns
- 2,020railway and bus stations
- 2,350places to stay
- 205lakes, waterfalls and rivers
- 1,204sights, museums and craft places
- 275plant and animal species
- 253dishes and products
Updated 11 November 2026 at 23:06.
What a graph is, in plain words
A graph stores knowledge as a network instead of a table. It has only two kinds of parts:
- Nodes — anything with a name and a place: a peak, a spring, a hut, a village, a trail, a railway station, a dish, a plant or an animal.
- Links — short three-part sentences: «trail — passes — Hoverla», «spring — is near — the summit», «banosh — comes from — Hutsulshchyna», «rhododendron — grows near — the path».
Think of a metro map: stations are nodes, the lines between them are links. We just have hundreds of thousands of «lines», and each one means something: near, passes, starts at, grows near, is cooked in.
Example: what the graph knows about Hoverla
Here is a real piece of the graph around the highest peak of Ukraine — exactly as the site sees it. The numbers are counted live from the same data as the pages.
Hoverla in the graph
Each circle is a group of neighbours Hoverla is linked to. Below are examples with links to their pages.
- 109 trails — pass the summit or lead to itZaroslyak base, via Mount Hoverla and Lake Nesamovyte to Zaroslyak base · Transcarpathian Tourist Trail · Kozmeshchyk – Mount Hoverla – Kozmeshchyk · Zaroslyak base – Mount Hoverla and 105 more
- 8 springs — near the summitSpring (Voronenko, 8) · Spring (Luhy, 2) · Spring below Mount Hoverla · Spring below Mount Hoverla (2) and 4 more
- 4 huts and shelters — nearbyEco point “Peremychka” (hut) · “Carpathian Highlands” information and tourist centre (shelter) · Picnic shelter below Mount Hoverla · Eco point “At the foot of Hoverla” (hut)
- 5 peaks — next to itMount Breskul · Mount Pozhyzhevska · Mount Velyka Kozmeska · Mala Kozmeska and 1 more
- 3 mountain meadows — nearbyPolonyna Pozhizhevska (Pozhizhevskaya) · Polonyna Skopeska · Polonyna Holovcheska
- 12 species — seen nearbyAlpine coltsfoot · Golden cinquefoil · Water pipit · Alpine bellflower and 8 more
- 4 starting points — where hikes beginLuhy · Pozhyzhevska · Zaroslyak · Vorokhta
- where it is — range, region, countryChornohora · Ivano-Frankivsk region · Ukraine
What it gives you
- The AI guide remembers the conversation. After a question about Hoverla, «what is nearby?» is answered from Hoverla's neighbours in the graph — not guessed.
- Search understands different names. For every place the graph keeps its spellings: the official name, the local one, Latin and common typos — «Говерла», «Hoverla», «Goverla».
- «Nearby» really means nearby. What to see around a guesthouse, the springs and huts on a trail, the plants and animals seen near a path — all come from the same links.
- Answers rely on the site's facts. Heights, distances and names come from the graph, so the AI guide makes things up less.
Where the data comes from
The graph is built automatically from what is already published on the site: the pages of peaks, trails, stays, sights, food and nature. Springs and huts come from OpenStreetMap, heights from the Copernicus elevation model, plant and animal records from open scientific databases. More on the Data & licences page.
As soon as an editor or a guesthouse owner changes a page, the graph updates its links within seconds, and every night it is rebuilt from scratch.
What the graph does not do
- It does not collect data about visitors: it holds only places and facts about them.
- It does not invent links: «near» is measured by coordinates, «passes» by the trail's track.
- It does not replace caution in the mountains: check the weather and the marking on the spot.
Questions and answers
Is the graph artificial intelligence?
No. The graph is organised data. The AI guide only uses it to answer more precisely.
Can I see the whole graph?
It is too big for one picture, but every page shows its own part: trails to a summit, springs on a trail, what is around a stay.
How do I get my place into the graph?
Add your place in the account — once published it joins the graph with all its links by itself.
