Use results to build model for finding errors
Bug #486420 reported by
Michael Wayne Goodman
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Egad |
Fix Committed
|
High
|
Michael Wayne Goodman |
Bug Description
Currently the analysis stage is mostly working and we are getting EgadItems generated, but not doing anything with them. The items should be fed into a model (use nltk's DecisionTree module first, possibly add more later (like the van Noord or de Kok style algorithms))
Changed in egad: | |
importance: | Undecided → High |
milestone: | none → 0.2.x |
assignee: | nobody → Michael Wayne Goodman (goodmami) |
To post a comment you must log in.
Fixed in r10.
The nltk's MaxentClassifier is currently implemented, along with a custom simple statistical classifier. DecisionTreeCla ssifier is halfway implemented, but it might not be useful for ranking (different than just classification).