Add min-cartesian-feature (as optional value)

src/GEval/BlackBoxDebugging.hs

@@ -6,5 +6,6 @@ data BlackBoxDebuggingOptions = BlackBoxDebuggingOptions {
   bbdoMinFrequency :: Integer,
   bbdoWordShapes :: Bool,
   bbdoBigrams :: Bool,
-  bbdoCartesian :: Bool
+  bbdoCartesian :: Bool,
+  bbdoMinCartesianFrequency :: Maybe Integer
 }

src/GEval/LineByLine.hs

@@ -32,6 +32,7 @@ import qualified Data.Conduit.Text as CT
 import Data.Text
 import Data.Text.Encoding
 import Data.Conduit.Rank
+import Data.Maybe (fromMaybe)
 
 import Data.List (sortBy, sort, concat)
 
@@ -129,7 +130,7 @@ formatFeatureWithPValue (FeatureWithPValue f p avg c) =
 
 featureExtractor :: Monad m => GEvalSpecification -> BlackBoxDebuggingOptions -> ConduitT (Double, LineRecord) RankedFeature m ()
 featureExtractor spec bbdo = CC.map extract
-                             .| finalFeatures (bbdoCartesian bbdo) (bbdoMinFrequency bbdo)
+                             .| finalFeatures (bbdoCartesian bbdo) (fromMaybe (bbdoMinFrequency bbdo) (bbdoMinCartesianFrequency bbdo))
                              .| CC.map unwrapFeatures
                              .| CC.concat
   where extract (rank, line@(LineRecord _ _ _ _ score)) =

src/GEval/OptionsParser.hs

@@ -165,13 +165,16 @@ specParser = GEvalSpecification
         )
       )
 
+defaultMinFrequency :: Integer
+defaultMinFrequency = 1
+
 blackBoxDebuggingOptionsParser :: Parser BlackBoxDebuggingOptions
 blackBoxDebuggingOptionsParser = BlackBoxDebuggingOptions
   <$> option auto
     ( long "min-frequency"
       <> metavar "N"
       <> help "Minimum frequency for the worst features"
-      <> value 1
+      <> value defaultMinFrequency
       <> showDefault)
   <*> switch
     ( long "word-shapes"
@@ -182,6 +185,10 @@ blackBoxDebuggingOptionsParser = BlackBoxDebuggingOptions
   <*> switch
     ( long "cartesian"
       <> help "Consider Cartesian combination of all features (computationally expensive!)")
+  <*> optional (option auto
+                ( long "min-cartesian-frequency"
+                  <> metavar "N"
+                  <> help "When combining features into Cartesian features, consider only features whose frequency exceeds the threshold given"))
 
 singletonMaybe :: Maybe a -> Maybe [a]
 singletonMaybe (Just x) = Just [x]