Add LC-QuAD_2 to the analysis part.

aorogat · Sep 15, 2021 · ceb693c · ceb693c
1 parent b7d7297
commit ceb693c
Show file tree

Hide file tree

Showing 4 changed files with 52 additions and 15 deletions.
diff --git a/src/DataSet/Benchmark.java b/src/DataSet/Benchmark.java
@@ -23,6 +23,7 @@ public class Benchmark {
 
     public final static int QALD_ALL = 100;
     public final static int LC_QUAD = 101;
+    public final static int LC_QUAD_2 = 111;
     public final static int GraphQuestions = 102;
     public final static int WebQuestions = 103;
     public final static int SimpleQuestions = 104;

diff --git a/src/DataSet/DataSetPreprocessing.java b/src/DataSet/DataSetPreprocessing.java
@@ -6,6 +6,7 @@
 import java.util.Scanner;
 import org.apache.jena.query.Query;
 import org.apache.jena.query.QueryFactory;
+import org.apache.jena.sys.JenaSystem;
 import qa.dataStructures.Question;
 import qa.parsers.JSONParser;
 import qa.parsers.Parser;
@@ -107,6 +108,14 @@ public static ArrayList getQueriesWithoutDuplicates(int benchmark) {
             questions.addAll(JSONParser.parseNo(currentDirectory+"/data/DBpedia/No_SPARQL/V1/test.json", "QUAD", "dbpedia"));
             questions.addAll(JSONParser.parseNo(currentDirectory+"/data/DBpedia/No_SPARQL/V1/valid.json", "QUAD", "dbpedia"));
         }
+
+        if (benchmark == Benchmark.LC_QUAD_2 || benchmark == Benchmark.PropertiesDefined) {
+            //questions.addAll(JSONParser.parseQuAD2File(currentDirectory + "/data/DBpedia/SPARQL/LC-QuAD2-data/test.json", "QUAD2", "wikidata"));
+            //questions.addAll(JSONParser.parseQuAD2File(currentDirectory + "/data/DBpedia/SPARQL/LC-QuAD2-data/train.json", "QUAD2", "wikidata"));
+            questions.addAll(JSONParser.parseQuAD2File(currentDirectory + "/data/DBpedia/SPARQL/LC-QuAD2-data/lcquad_2_0.json", "QUAD2", "wikidata"));
+
+        }
+
         if (benchmark == Benchmark.GraphQuestions || benchmark == Benchmark.PropertiesDefined) {
             questions.addAll(JSONParser.parseGra(currentDirectory + "/data/Freebase/SPARQL/GraphQuestions-master/freebase13/graphquestions.testing.json", "Freebase_Graph", "freebase"));
             questions.addAll(JSONParser.parseGra(currentDirectory + "/data/Freebase/SPARQL/GraphQuestions-master/freebase13/graphquestions.training.json", "Freebase_Graph", "freebase"));
@@ -219,6 +228,8 @@ public static ArrayList getQueriesWithoutDuplicates(int benchmark) {
          Note that even with this turned on there are still lots of non-standard SPARQL syntactic 
          constructs that Virtuoso supports that ARQ will still reject.
          */
+        JenaSystem.init();
+        org.apache.jena.query.ARQ.init();
         for (int i = 0; i < questions.size(); i++) {
             String queryString = questions.get(i).getQuestionQuery();
             try {

diff --git a/src/mainClass/BenchmarkAnalysis.java b/src/mainClass/BenchmarkAnalysis.java
@@ -27,7 +27,7 @@ public static void analyze()
         System.out.println("                 11- LC-QUAD, \t\t12- WebQuestions, \t\t13- GraphQuestions, ");
         System.out.println("                 14- ComplexQuestions, \t15- ComQA, \t\t\t16- TempQuestions, ");
         System.out.println("                 17- SimpleDBpediaQA, \t18- SimpleQuestions, ");
-        System.out.println("                 19- User Defined]");
+        System.out.println("                 19- User Defined, \t20- LC-QuAD_2]");
         System.out.print("               Benchmark is: ");
 
         benchmark = in.nextInt();
@@ -53,6 +53,7 @@ public static void analyze()
            case 17: benchmark = Benchmark.SimpleDBpediaQA; break;
            case 18: benchmark = Benchmark.SimpleQuestions; break;
            case 19: benchmark = Benchmark.UserDefined; break;
+           case 20: benchmark = Benchmark.LC_QUAD_2; break;
            default: analyze();
        }
 
@@ -62,20 +63,20 @@ public static void analyze()
         //Shallow Anlysis
         //===============
         //1- Keyword Analysis
-        Keywords k = new Keywords(qs);
-        k.keywordsAnalysis();
-
-        //2- Number of triples analysis
-        NoOfTriples triples = new NoOfTriples(qs);
-        triples.triplesAnalysis();
-
-        //3- Operator Distribution
-        OperatorDistribution distribution = new OperatorDistribution(qs);
-        distribution.analysis();
-
-        //Shape Analysis
-        ShapeAnalysis shapes = new ShapeAnalysis(qs);
-        shapes.analysis();
+//        Keywords k = new Keywords(qs);
+//        k.keywordsAnalysis();
+//        
+//        //2- Number of triples analysis
+//        NoOfTriples triples = new NoOfTriples(qs);
+//        triples.triplesAnalysis();
+//        
+//        //3- Operator Distribution
+//        OperatorDistribution distribution = new OperatorDistribution(qs);
+//        distribution.analysis();
+//        
+//        //Shape Analysis
+//        ShapeAnalysis shapes = new ShapeAnalysis(qs);
+//        shapes.analysis();
 
         //NLQ Analysis
         QABenchmark.NLQAnalyze();

diff --git a/src/qa/parsers/JSONParser.java b/src/qa/parsers/JSONParser.java
@@ -163,6 +163,30 @@ public static ArrayList<Question> parseQuADFile(String fileDirectory, String sou
         }
         return questionsList;
     }
+
+    public static ArrayList<Question> parseQuAD2File(String fileDirectory, String sourceString, String endpoint) throws JSONException {
+        ArrayList<Question> questionsList = new ArrayList<Question>();
+        String fileContents = FileManager.readWholeFile(fileDirectory);
+        JSONArray arr = new JSONArray(fileContents);
+        for (int i = 0; i < arr.length(); i++) {
+
+            Question question = new Question();
+            question.setDatabase(endpoint);
+            question.setFilepath(fileDirectory);
+            // The source is predefined
+            question.setQuestionSource(sourceString);
+
+            JSONObject currentQuestionObject = arr.getJSONObject(i);
+            question.setQuestionString(currentQuestionObject.get("paraphrased_question").toString().replace("\n", ""));
+
+            // The question's SPARQL Query
+            question.setQuestionQuery(currentQuestionObject.get("sparql_wikidata").toString());
+            if (question.getQuestionQuery() != null || question.getQuestionString()!=null && !question.getQuestionString().equals("")) {
+                questionsList.add(question);
+            }
+        }
+        return questionsList;
+    }
 
     public static ArrayList<Question> parseNo(String fileDirectory, String sourceString, String endpoint) throws JSONException {
         ArrayList<Question> questionsList = new ArrayList<Question>();