Page Function Classifier for Vilnius Clojure Dojo

Code ready for modification to increase page classifier accuracy and precision.

Usage

Unzip labeled data

unzip dataset.zip

Generate page features

lein generate-features features.tsv

Cross-validate classifier

lein cross-validate features.tsv

Train classifier

lein train features.tsv

Evaluate classifier (optional, do not use data used for training)

lein evaluate test.tsv

Test classifier with real page

lein classify "http://..."

Add page to dataset (optional)

lein add-page "L" "http://..."

Use different classifier: src/pageclass/train.clj Available ones extend AbstractClassifier.
Generate meaningful features: src/pageclass/features.clj