Closed alexandermorgan closed 9 years ago
This looks like a good opportunity to simplify how settings are passed to the interval indexer's "real_indexer_func".
Could you please put this in the vis-rodan github? It's not a client issue but a wrapper issue. Tx.
On Wed, Aug 26, 2015 at 3:40 PM, Alexander Morgan notifications@github.com wrote:
This looks like a good opportunity to simply how settings are passed to the interval indexer's "real_indexer_func".
— Reply to this email directly or view it on GitHub https://github.com/ELVIS-Project/vis-framework/issues/361#issuecomment-135148993 .
VIS used to have this capability, but I don't think it ever made it into the Framework era. Not sure what Ryan means—this definitely seems like a setting for the IntervalIndexer
.
I agree that this should be in VIS itself because this will change the way the indexer_func in interval.py will work.
There are currently settings to allow for analysis with or without interval quality and with simple or compound intervals, but a new setting is needed to provide absolute intervals (referred to as "canonical" intervals in other closed issues pertaining to the n-gram indexer). This would combine, for example, P4s and -P4s, meaning that the count would not be effected by which voice is on top. While less pressing, this may be a desirable setting to add tot he horizontal interval indexer too. These must be user-controllable settings because different queries will require different settings. The default should be kept as it is, where uncrossed and crossed versions of intervals are counted separately.