Porting Semantic Role Labeling engine from v1 and add unit test cover…

…age #4
Lambda-3 · May 10, 2017 · dd42441 · dd42441
1 parent 425cb2f
commit dd42441
Show file tree

Hide file tree

Showing 3 changed files with 48 additions and 19 deletions.
diff --git a/stargraph-core/src/main/java/net/stargraph/core/query/QueryEngine.java b/stargraph-core/src/main/java/net/stargraph/core/query/QueryEngine.java
@@ -176,7 +176,7 @@ private void resolvePredicate(InstanceEntity pivot, DataModelBinding binding, SP
             ModifiableSearchParams searchParams = ModifiableSearchParams.create(dbId).term(binding.getTerm());
             ModifiableRankParams rankParams = ParamsBuilder.word2vec();
             Scores scores = searcher.pivotedSearch(pivot, searchParams, rankParams);
-            builder.add(binding, scores.stream().limit(3).collect(Collectors.toList()));
+            builder.add(binding, scores.stream().limit(6).collect(Collectors.toList()));
         }
     }
 

diff --git a/stargraph-core/src/main/java/net/stargraph/core/query/nli/AnalysisStep.java b/stargraph-core/src/main/java/net/stargraph/core/query/nli/AnalysisStep.java
@@ -98,12 +98,16 @@ AnalysisStep resolve(DataModelTypePattern rule) {
 
         if (matches(rulePattern)) {
             String newQuestionStr;
-            String newPosTagStr = posTagStr;
+            String newPosTagStr;
 
             if (rule.isLexical()) {
                 String placeHolder = createPlaceholder(questionStr, modelType);
                 Replacement replacement = replace(rulePattern, questionStr, placeHolder);
                 newQuestionStr = replacement.value;
+
+                String subPosStr = findSubPosStr(replacement);
+                Replacement posTagReplacement = replace(subPosStr, posTagStr, placeHolder);
+                newPosTagStr = posTagReplacement.value;
             }
             else {
                 String placeHolder = createPlaceholder(posTagStr, modelType);
@@ -145,6 +149,24 @@ private String findSubStr(Replacement replacement) {
         return subStr;
     }
 
+    private String findSubPosStr(Replacement replacement) {
+        String[] capture = Objects.requireNonNull(replacement).capture.split("\\s");
+        int startIdx = 0;
+        String subPosStr = null;
+        for (Word w : annotated) {
+            if (w.getText().equals(capture[0])) {
+                subPosStr = annotated.stream()
+                        .skip(startIdx)
+                        .limit(capture.length)
+                        .map(word -> word.getPosTag().getTag()).collect(Collectors.joining(" "));
+                break;
+            }
+            startIdx++;
+        }
+
+        return subPosStr;
+    }
+
     private boolean matches(Pattern pattern) {
         Matcher m1 = pattern.matcher(posTagStr);
         Matcher m2 = pattern.matcher(questionStr);

diff --git a/stargraph-core/src/main/resources/reference.conf b/stargraph-core/src/main/resources/reference.conf
@@ -151,7 +151,8 @@ stargraph.rules = {
     ".* (their) .*",
     ".* (the) .*",
     ".* (of) .*",
-    ".*(TYPE_1).*"
+    ".*(TYPE_1).*",
+    ".*(STOP_1).*"
   ]
 
   syntatic-pattern.en = [
@@ -162,24 +163,30 @@ stargraph.rules = {
     #
     # The order is relevant. Split the array if need to interleave rules.
 
+    {
+      STOP: [
+        ".* (did) .*"
+      ]
+    }
+
     {
       OPERATION: [
-        "same type as",
-        "same",
-        "second highest",
-        "largest",
-        "highest after",
-        "first",
-        "more than",
-        "less than",
-        "earlier than",
-        "between",
-        "more",
-        "less",
-        "most",
-        "past",
-        "still",
-        "also"
+        ".* (same type as) .*",
+        ".* (same) .*",
+        ".* (second highest) .*",
+        ".* (largest) .*",
+        ".* (highest after) .*",
+        ".* (first) .*",
+        ".* (more than) .*",
+        ".* (less than) .*",
+        ".* (earlier than) .*",
+        ".* (between) .*",
+        ".* (more) .*",
+        ".* (less) .*",
+        ".* (most) .*",
+        ".* (past) .*",
+        ".* (still) .*",
+        ".* (also) .*"
       ]
     }