number of words used for creating patterns can now be adjusted (indep…

…endent of context size)

src/main/java/io/github/infolis/algorithm/FrequencyBasedBootstrapping.java

@@ -34,7 +34,7 @@ public FrequencyBasedBootstrapping(DataStoreClient inputDataStoreClient, DataSto
 	private static final org.slf4j.Logger log = LoggerFactory.getLogger(FrequencyBasedBootstrapping.class);
 
 	public PatternInducer getPatternInducer() {
-	   	return new StandardPatternInducer();
+	   	return new StandardPatternInducer(getExecution().getWindowsize());
 	}
 
 	public PatternRanker getPatternRanker() {

src/main/java/io/github/infolis/algorithm/ReliabilityBasedBootstrapping.java

@@ -40,7 +40,7 @@ public ReliabilityBasedBootstrapping(DataStoreClient inputDataStoreClient, DataS
     private Reliability r = new Reliability();
 
     public PatternInducer getPatternInducer() {
-    	return new StandardPatternInducer();
+    	return new StandardPatternInducer(getExecution().getWindowsize());
     }
 
     public PatternRanker getPatternRanker() {

src/main/java/io/github/infolis/infolink/patternLearner/StandardPatternInducer.java

@@ -30,12 +30,16 @@
 public class StandardPatternInducer extends Bootstrapping.PatternInducer {
 
 	private static final org.slf4j.Logger log = LoggerFactory.getLogger(StandardPatternInducer.class);
-	// TODO derive from windowsize
-	public final static int patternsPerContext = 9;
+
+    private int windowsize;;
+	private static int patternsPerContext;
 	Pattern leadingWildcards = Pattern.compile("\"(\\*\\s)+");
 	Pattern trailingWildcards = Pattern.compile("(\\s\\*)+\"");
 
-	public StandardPatternInducer() {}
+	public StandardPatternInducer(int windowsize) {
+		this.windowsize = windowsize;
+		patternsPerContext = (windowsize * 2) -1;
+	}
 
 	public final int getPatternsPerContext() {
 		return patternsPerContext;
@@ -104,8 +108,6 @@ public String apply(String s) {
         String delimiter_left = leftWords.get(0);
         String delimiter_right = rightWords.get(0);
 
-        // TODO make configurable
-        int windowsize = 5;
         // set default values in case the context of a term contains less elements than the given windowsize
         List<InfolisPattern> inducedPatternsLeft = Stream.generate(InfolisPattern::new)
         												.limit(windowsize)
@@ -150,7 +152,13 @@ public String apply(String s) {
         }
 	    // order is important here: patterns are listed in ascending order with regard to their generality
 	    // type2left and type2right etc. have equal generality
-	    return (Arrays.asList(typeGeneral, inducedPatternsLeft.get(0), inducedPatternsRight.get(0), inducedPatternsLeft.get(1), inducedPatternsRight.get(1), inducedPatternsLeft.get(2), inducedPatternsRight.get(2), inducedPatternsLeft.get(3), inducedPatternsRight.get(3)));
+        List<InfolisPattern> patterns = new ArrayList<>();
+        patterns.add(typeGeneral);
+        for (int i = 0; i < windowsize -1; i++) {
+        	patterns.add(inducedPatternsLeft.get(i));
+        	patterns.add(inducedPatternsRight.get(i));
+        }
+        return patterns;
 	}
 
 }

src/main/java/io/github/infolis/model/Execution.java

@@ -243,23 +243,33 @@ public Algorithm instantiateAlgorithm(
 
 	/**
          * Any kind of search query that can be used within the algorithms.
-         * For example, it represtens the search query which is used
+         * For example, it represents the search query which is used
          * to perform a search in different repositories to find
          * fitting research data.
          * 
-	 * {@link LuceneSearcher} {@link FederatedSearcher} {@link ApplyPatternAndResolve}
+	 * {@link FederatedSearcher} {@link SearchPatternsAndCreateLinks}
 	 */
 	private String searchQuery;
 
 	/**
-	 * Group numbers to use for RegexSearcher.
+	 * Group numbers to use for RegexSearcher: group of reference term.
 	 * 
 	 * {@Link RegexSearcher}
 	 */
 	private int referenceGroup = RegexUtils.doiGroupNum;
 
+	/**
+	 * Group numbers to use for RegexSearcher: group of left context.
+	 * 
+	 * {@Link RegexSearcher}
+	 */
 	private int leftContextGroup = RegexUtils.doiLeftContextGroupNum;
 
+	/**
+	 * Group numbers to use for RegexSearcher: group of right context.
+	 * 
+	 * {@Link RegexSearcher}
+	 */
 	private int rightContextGroup = RegexUtils.doiRightContextGroupNum;
 
 	/**
@@ -309,6 +319,13 @@ public Algorithm instantiateAlgorithm(
          * {@link Bootstrapping}
 	 */
 	private int maxIterations = 10;
+
+	/**
+	 * Number of words used for creation of patterns.
+	 * 
+	 * {@link StandardPatternInducer}
+	 */
+	private int windowsize = 3;
 
 
         //TODO: also used for frequencyBasedBootstrapping, should we just name 
@@ -682,6 +699,14 @@ public int getMaxIterations() {
 	public void setMaxIterations(int maxIterations) {
 		this.maxIterations = maxIterations;
 	}
+
+	public int getWindowsize() {
+		return this.windowsize;
+	}
+
+	public void setWindowsize(int windowsize) {
+		this.windowsize = windowsize;
+	}
 
 	public double getReliabilityThreshold() {
 		return reliabilityThreshold;

src/test/java/io/github/infolis/infolink/patternLearner/StandardPatternInducerTest.java

@@ -27,7 +27,7 @@ public class StandardPatternInducerTest {
 
 	@Test
 	public void testInduce() {
-		StandardPatternInducer inducer = new StandardPatternInducer();
+		StandardPatternInducer inducer = new StandardPatternInducer(5);
 		Double[] thresholds = {0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0};
 
 		TextualReference ref = new TextualReference("15757 41727 5743 10877 10014 30850 Sozialstaatssurvey/", "ALLBUS", " .", "textfile", "pattern", "mentionsReference");