Add smart mode

src/GEval/MatchingSpecification.hs

@@ -18,6 +18,9 @@ module GEval.MatchingSpecification
 import Data.Singletons.TH
 import Data.Text
 import Data.List.Extra (breakOn)
+import Data.Char (isLetter)
+import Data.List (find)
+import Data.Maybe (isJust)
 
 import Text.EditDistance
 
@@ -26,6 +29,8 @@ singletons [d|data MatchingSpecification = ExactMatch -- ^ exact match, i.e. ide
                                          | FuzzyMatch -- ^ fuzzy match by Levenshtein distance
                                          | CutLabel MatchingSpecification -- ^ require that the label (part before up to `=`)
                                                                           -- is matched and then proceed with some matching spec.
+                                         | SmartMatch MatchingSpecification -- ^ do fuzzy matching only on values
+                                                                            -- containing letters
                                          deriving (Eq)
              |]
 
@@ -41,6 +46,16 @@ getMatchingFunctionForString (CutLabel smatchSpec) a b = getMatchingFunctionForS
   where a' = cutLabel a
         b' = cutLabel b
 
+getMatchingFunctionForString (SmartMatch smatchSpec) got expected = getMatchingFunctionForString chosenMatch got expected
+  where chosenMatch = if wantedBySmartMatch expected
+                      then smatchSpec
+                      else ExactMatch
+
+-- | Whether suitable for fuzzy matching when in the "smart" match mode.
+-- At the moment we check whether it contains at least one letter
+-- (we require the exact match for, for instance, numbers written with digits.
+wantedBySmartMatch = isJust . (Data.List.find isLetter)
+
 -- | Remove the label along with the separator (the equal sign)
 cutLabel :: String -> String
 cutLabel t = case Data.List.Extra.breakOn "=" t of

src/GEval/Metric.hs

@@ -83,6 +83,7 @@ instance Show Metric where
   show (MultiLabelFMeasure beta ExactMatch) = "MultiLabel-F" ++ (show beta)
   show (MultiLabelFMeasure beta FuzzyMatch) = "Fuzzy/" ++ (show $ MultiLabelFMeasure beta ExactMatch)
   show (MultiLabelFMeasure beta (CutLabel matchSpec)) = "CutLabel/" ++ (show $ MultiLabelFMeasure beta matchSpec)
+  show (MultiLabelFMeasure beta (SmartMatch matchSpec)) = "Smart/" ++ (show $ MultiLabelFMeasure beta matchSpec)
   show MultiLabelLogLoss = "MultiLabel-Logloss"
   show MultiLabelLikelihood = "MultiLabel-Likelihood"
   show (Mean metric) = "Mean/" ++ (show metric)
@@ -104,6 +105,9 @@ instance Read Metric where
   readsPrec p ('C':'u':'t':'L':'a':'b':'e':'l':'/':theRest) = case readsPrec p theRest of
     [(metric, theRest)] -> [(applyMatchingSpecification CutLabel metric, theRest)]
     _ -> []
+  readsPrec p ('S':'m':'a':'r':'t':'/':theRest) = case readsPrec p theRest of
+    [(metric, theRest)] -> [(applyMatchingSpecification SmartMatch metric, theRest)]
+    _ -> []
   readsPrec _ ('R':'M':'S':'E':theRest) = [(RMSE, theRest)]
   readsPrec _ ('M':'S':'E':theRest) = [(MSE, theRest)]
   readsPrec _ ('P':'e':'a':'r':'s':'o':'n':theRest) = [(Pearson, theRest)]

test/Spec.hs

@@ -349,6 +349,8 @@ main = hspec $ do
       runGEvalTest "multilabel-f1-ie-flags" `shouldReturnAlmost` 0.444444444444
     it "information extraction with fuzzy matching" $ do
       runGEvalTest "multilabel-f1-ie-fuzzy" `shouldReturnAlmost` 0.681777777777
+    it "information extraction with smart fuzzy matching" $ do
+      runGEvalTest "multilabel-f1-ie-fuzzy-smart" `shouldReturnAlmost` 0.598444
   describe "Mean/MultiLabel-F" $ do
     it "simple" $ do
       runGEvalTest "mean-multilabel-f1-simple" `shouldReturnAlmost` 0.5

test/multilabel-f1-ie-fuzzy-smart/multilabel-f1-ie-fuzzy-smart-solution/test-A/out.tsv

@@ -0,0 +1,3 @@
+important-person=JOHN_BROWN important-person=JOHN_SMITH company-name=Axaxaxaxas_Mlo profit=12031
+company-name=Foo_Bar profit=1220
+company-name=Whatever important-person=PIERRE_MENARD

test/multilabel-f1-ie-fuzzy-smart/multilabel-f1-ie-fuzzy-smart/config.txt

@@ -0,0 +1 @@
+--metric CutLabel/Smart/Fuzzy/MultiLabel-F1:ls<_(inc|ltd)\.?(\s|$)><\2>

test/multilabel-f1-ie-fuzzy-smart/multilabel-f1-ie-fuzzy-smart/test-A/expected.tsv

@@ -0,0 +1,3 @@
+company-name=Axaxaxas_Mlö profit=12031 important-person=John_Smith important-person=James_Brown
+company-name=Orbis_Tertius profit=1020 important-person=Anna_Smith
+company-name=Whatever_Inc profit=5600 important-person=Pierre_Menard