This directory contains data used for testing Spiral's ability to split identifiers. It currently consists of two "oracle" files (i.e., sets of identifiers together with their splits, based on human or other validation):
- The Loyola University of Delaware Identifier Splitting Oracle (Ludiso). The files came from this website as it existed on 2017-02-18: http://www.cs.loyola.edu/~binkley/ludiso/
- The INTT data set, extracted from the zip archive of INTT put online in 2011 at the following website: http://oro.open.ac.uk/28352/
The files in this directory are in tab-separated (TSV) form. Each row has two columns:
identifer each,token,split,separated,by,commas
The second column shows the split version of the identifier with each token separated by commas. Here is an example:
windowspan window,span
The archived subdirectory contains archived versions of the original materials from which the .tsv files in this directory were created.