What Korea’s Barrier-Free Data Records

9,677 venues, 28 possible details

Korea’s national tourism organisation publishes barrier-free accessibility information for tourist venues — museums, parks, temples, markets, observation decks. Our catalogue holds 9,677 of them. Each can record any combination of 28 different accessibility features.

8,840 venues (91.4%) record at least one feature. The remaining 837 appear in the barrier-free dataset with no accessibility details at all — listed as a venue, described as nothing. That is the first thing worth knowing about this data.

The second is what the 28 fields are actually used for. Grouped by who each feature serves, the distribution is not close.

Mobility is documented 172 times more often than hearing

32,352 against 188. Mobility features are recorded 172 times more often than features for deaf and hard-of-hearing visitors. Per field the gap is 2,941 against 38 — a factor of 78.

The single sharpest comparison is between the most-recorded and least-recorded fields. Accessible routes are documented at 7,115 venues. Sign language assistance is documented at 12. That is a ratio of 593 to 1, inside one dataset, collected by one organisation, on one form.

All 28 fields, in order

The shape of that list is a long head and a very long tail. The top five fields account for 26,693 of the 41,150 total entries — 65%. The bottom ten account for 961 between them, 2.3%.

What the imbalance does and does not tell you

The honest reading is narrower than the striking numbers invite, and it is worth being explicit about the limits before drawing any conclusion.

This measures documentation, not provision. A venue with no sign-language entry may still have staff who sign. A venue with no lift entry may have three lifts. What the data captures is which questions Korean venues have been asked, and have answered.

That said, the pattern is too lopsided to be purely reporting noise. Accessible parking and lifts are physical installations that cost money and get built because regulation requires them; they also generate paperwork. Sign language assistance is a staffing arrangement, not a structure — it is less visible, less regulated, and easier to leave unrecorded.

Korean accessibility policy has historically been strongest on the built environment. Ramps, lifts, tactile paving and accessible parking are all covered by construction and facility standards with inspection regimes behind them. Communication access has no equivalent. The dataset reflects that difference faithfully, which is arguably its most useful property.

What the form does not ask

A survey instrument defines what can be known. Reading the 28 fields as a list of questions, several things are conspicuously absent — and their absence is invisible unless you go looking, because a field that does not exist cannot show up as empty.

• Nothing about gradient or distance. “Accessible route” is a yes/no. A 400-metre ramped path at 1:12 and a flat 20-metre entrance record identically, and the difference decides whether a manual wheelchair user can make the trip.
• Nothing about sensory environment. No field for quiet spaces, lighting, or noise — the provisions autistic visitors and people with sensory sensitivities most often need.
• Nothing about cognitive accessibility. No easy-read materials, no plain-language signage, no wayfinding simplicity.
• Nothing about staff training, beyond a single yes/no for whether guide staff exist.
• Nothing about when. Every field is a permanent property. A lift that is out of service for six months looks the same as one that works.

None of this makes the dataset bad. It makes it a product of when it was designed — an era when accessibility policy meant the built environment. The 593-to-1 ratio between accessible routes and sign language is the same story told from the other end.

It is also worth knowing how much of Korea this covers. The barrier-free programme assesses 9,677 venues. The general tourism dataset we hold alongside it lists 48,747 places. We have not matched the two records-by-record, so we cannot tell you the exact overlap — but the barrier-free set is unambiguously a selected subset, not a survey of everywhere you might want to go. A venue absent from it has not failed an assessment. It has not had one.

Why we publish the fields rather than a score

It would be straightforward to reduce these 28 fields to a single “accessibility rating” per venue and sort by it. We do not, and this dataset is the clearest argument for why not.

• A score would rank a venue with a lift above a venue with a sign-language interpreter, because lifts are recorded 242 times more often. That ranks the paperwork, not the place.
• A score hides which access need is met. “Accessible: 7/10” is useless if you need the three that are missing.
• A score turns unrecorded into a penalty. With 837 venues recording nothing, any average would push them to the bottom of a list where they may not belong.
• Different visitors need different fields. The relevant question is never “how accessible”, it is “accessible to whom”.

So we show what the source records, field by field, and label absent data as absent. It makes for a messier interface than a badge would. It is the only version that stays true when you look closely.

If you take one thing from this article, make it the ratio. 7,115 to 12. One dataset, one form, one organisation — and a six-hundred-fold difference in how thoroughly two kinds of visitor are described. That gap was not decided by any venue. It was decided by which questions got asked, and by which of them anyone was ever required to answer.

Sources and method

• Source: 한국관광공사_무장애 여행 정보 — Korea Tourism Organization barrier-free travel information, published on Korea’s public data portal.
• Counts taken from our catalogue on 6 August 2026: 9,677 venues, 8,840 with at least one accessibility field recorded, 41,150 field entries in total across the 28 fields.
• A venue counts toward a feature only where that feature carries a non-empty value. Field groupings are ours, based on the dataset’s own field definitions.