In uk.transport.london Jon Passenger wrote:
In the current climate, the application of some basic knowledge to
create rule of thumb heuristics (ie test splitting at major
interchanges / regional boundaries / known price break points ) plus
attempts at some sort of primitive 'crowdsourcing' (yuk) - like the
site cited above - might be the best available approach.

I think a cross between Skyscanner and Property Bee might work. That is,
when users query the database for fares in the normal course of events, our
website somehow intercepts the query (in the Skyscanner case by doing it for
them, in the Property Bee case by intercepting their browser session as the
data goes past). As well as telling them the answer, the site also records
it. That means it builds up a partial database of results that it can
search quickly, and just need fire off queries for results that aren't
known. Provide a compelling interface to search for normal fares (maybe
something like SkyScanner or traintimes.org.uk) and you'll get enough
traffic to keep the database fresh.

How they get these past the 'database right' lawyers is another question,
though.

Theo