search

hcouch21 / styloproject

Automatically exported from code.google.com/p/styloproject
0 stars 0 forks source link

Need a way to detect outdated training data. #3

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
What steps will reproduce the problem?
1. Train a corpus with a set of features.
2. Classify a document with the same set of features. (Success)
3. Classify a document with a new set of features (Failure)

What is the expected output? What do you see instead?
- Stylo should tell the user the training data does not match, instead we get 
an exception.

Original issue reported on code.google.com by Matthew.Tornetta@gmail.com on 28 Feb 2011 at 7:32

GoogleCodeExporter commented 9 years ago 
This should be fixed as of revision 46f383d0e981.

Stylo will give an error message and gracefully exit if classification is done 
using a different feature set than what training was done on.  This uses a 
comment in the ARFF file, so it will reject any training done before this 
revision.

Original comment by ssbahe...@gmail.com on 2 May 2011 at 5:32