Prototype result classifier

[ravenscroftj]

To implement a result classifier:

• The best features for classifying papers by result need to be identified and used to build an example classifier
• A sample of papers must be taken from the partridge database for testing purposes
• An ML system should be taught about positive and negative results
• The system should be tested on some of the sample data and retrained accordingly
• Once a good accuracy is obtained, all papers in Partridge should be classified and the information added to the database

[ravenscroftj]

Since it seems to make more sense to try and build a paper type classifier first, this is being pushed back to iteration 3