From 07ab904d609379c93598b2b16f501964decc16dd Mon Sep 17 00:00:00 2001
From: miconis <m.debonis991@gmail.com>
Date: Thu, 16 Jul 2020 18:57:55 +0200
Subject: [PATCH] implementation of the clustering function for the
 suffixprefix chain

---
 .../WordsStatsSuffixPrefixChain.java          | 90 +++++++++++++++++++
 .../clustering/ClusteringFunctionTest.java    | 24 +++++
 2 files changed, 114 insertions(+)
 create mode 100644 dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsStatsSuffixPrefixChain.java

diff --git a/dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsStatsSuffixPrefixChain.java b/dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsStatsSuffixPrefixChain.java
new file mode 100644
index 000000000..6fa2668fa
--- /dev/null
+++ b/dnet-pace-core/src/main/java/eu/dnetlib/pace/clustering/WordsStatsSuffixPrefixChain.java
@@ -0,0 +1,90 @@
+package eu.dnetlib.pace.clustering;
+
+import com.google.common.collect.Sets;
+import eu.dnetlib.pace.config.Config;
+
+import java.util.*;
+import java.util.stream.Collectors;
+
+@ClusteringClass("wordsStatsSuffixPrefixChain")
+public class WordsStatsSuffixPrefixChain extends AbstractClusteringFunction {
+
+        public WordsStatsSuffixPrefixChain(Map<String, Integer> params) {
+            super(params);
+        }
+
+        @Override
+        protected Collection<String> doApply(Config conf, String s) {
+            return suffixPrefixChain(s, param("mod"));
+        }
+
+        private Collection<String> suffixPrefixChain(String s, int mod) {
+
+            //create the list of words from the string (remove short words)
+            List<String> wordsList =
+                    Arrays.stream(s.split(" "))
+                    .filter(si -> si.length() > 3)
+                    .collect(Collectors.toList());
+
+            final int words = wordsList.size();
+            final int letters = s.length();
+
+            //create the prefix: number of words + number of letters/mod
+            String prefix = words + "-" + letters/mod + "-";
+
+            return doSuffixPrefixChain(wordsList, prefix);
+
+        }
+
+        private Collection<String> doSuffixPrefixChain(List<String> wordsList, String prefix) {
+
+            Set<String> set = Sets.newLinkedHashSet();
+            switch(wordsList.size()){
+                case 0:
+                case 1:
+                    break;
+                case 2:
+                    set.add(
+                            prefix +
+                            suffix(wordsList.get(0), 3) +
+                            prefix(wordsList.get(1), 3)
+                    );
+
+                    set.add(
+                            prefix +
+                            prefix(wordsList.get(0), 3) +
+                            suffix(wordsList.get(1), 3)
+                    );
+
+                    break;
+                default:
+                    set.add(
+                            prefix +
+                            suffix(wordsList.get(0), 3) +
+                            prefix(wordsList.get(1), 3) +
+                            suffix(wordsList.get(2), 3)
+                    );
+
+                    set.add(
+                            prefix +
+                            prefix(wordsList.get(0), 3) +
+                            suffix(wordsList.get(1), 3) +
+                            prefix(wordsList.get(2), 3)
+                    );
+                    break;
+            }
+
+            return set;
+
+        }
+
+
+        private String suffix(String s, int len) {
+            return s.substring(s.length()-len);
+        }
+
+        private String prefix(String s, int len) {
+            return s.substring(0, len);
+        }
+
+}
diff --git a/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java b/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
index 832ba4bce..396604cea 100644
--- a/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
+++ b/dnet-pace-core/src/test/java/eu/dnetlib/pace/clustering/ClusteringFunctionTest.java
@@ -111,6 +111,30 @@ public class ClusteringFunctionTest extends AbstractPaceTest {
 		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
 	}
 
+	@Test
+	public void testWordsStatsSuffixPrefix() {
+		params.put("mod", 10);
+
+		final ClusteringFunction sp = new WordsStatsSuffixPrefixChain(params);
+
+		String s = "Search for the Standard Model Higgs Boson";
+		System.out.println(s);
+		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
+
+		s = "A Physical Education Teacher Is Like...: Examining Turkish Students  Perceptions of Physical Education Teachers Through Metaphor Analysis";
+		System.out.println(s);
+		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
+
+		s = "Structure of a Eukaryotic Nonribosomal Peptide Synthetase Adenylation Domain That Activates a Large Hydroxamate Amino Acid in Siderophore Biosynthesis";
+		System.out.println(s);
+		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
+
+		s = "Performance Evaluation";
+		System.out.println(s);
+		System.out.println(sp.apply(conf, Lists.newArrayList(title(s))));
+
+	}
+
 	@Test
 	public void testFieldValue() {