-
Notifications
You must be signed in to change notification settings - Fork 1
Testing
The indicator that relies on infobox template references is the most reliable indicator. We use this observation by determining three sets of infoboxes:
- The positive set contains all template names that give a clear indication on a software language knowledge trace.
- The negative set contains all template names that do not appear in an article with software language knowledge.
- The maybe set does not give a clear indication as names might appear in articles with software language knowledge and in articles without software language knowledge.
The positive infobox template references are:
The set of non-positive infobox template references is much larger. We explored all articles up to a depth of 8 under our chosen root categories and found 684 infobox template references that are not positive. We save each template reference together with its article names where they appear at https://github.com/softlang/wikionto/blob/master/data/seed_neg_pre.json .
For each non-positive infobox template reference, we pick a single random article and manually classify it as providing software language knowledge or not. This manual annotation is persisted at https://github.com/softlang/wikionto/blob/master/data/seed_neg.json
We gain the following infobox template references that are just sometimes appear in software language articles. In most cases, the article described an entity that could be interpreted as a language, but this interpretation is not obvious and has to be reviewed by many more experts to reach consensus.
- Sometimes now template was referenced.
- See Open Document
- unit
- See Byte
- data structure
- See Scapegoat Tree
- writing system
- See Nemeth Braille
- quality tool
- See Ishikawa diagram
- character encoding
- See ISO/IEC_8859
- identifier
- See Bibcode
From the exploration of the seed set, we already knew the following references that appear in software language articles.
- software
- technology standard
- software license
We systematically compare the results of single indicators.