Incorrect matches of short segments at some three way intersections?

[mharnold]

Paul Cone noted some 'false positives' in matching OSM to RLIS in the Portland area that suggest that the spiderosm core algorithm is failing to properly match intersections in these cases?

Adding images of example cases as separate comments.

[mharnold]

CASE 1

[mharnold]

CASE 2

[mharnold]

CASE 3

[mharnold]

CASE 2 & 3 may be hard to fix, since the underlying topology in RLIS (the jurisdictional data) is different than in OSM. In both cases two OSM intersections are represented as a single intersection in RLIS.

CASE 1 - Looks fixable. It may be that the distance between the intersection locations (coordinates) in the OSM vs RLIS exceeds the threshold at which spiderOSM will match them?