summarization / nicolas

Commit 1b04826e9c8d19c7e829cb9b9261112b8918d6b3

Authored by Mateusz Kopeć
1 parent c9b1021d

release

Inline Side-by-side
Showing 3 changed files with 14 additions and 0 deletions
nicolas-train/src/main/java/pl/waw/ipipan/zil/summ/nicolas/eval/SummarizeTestCorpus.java
  View file @1b04826
@@ -57,7 +57,11 @@ public class SummarizeTestCorpus { @@ -57,7 +57,11 @@ public class SummarizeTestCorpus {
57 57
58 private static Map<String, String> summarizeTexts(Map<String, TText> id2preprocessedText, Nicolas nicolas) throws NicolasException { 58 private static Map<String, String> summarizeTexts(Map<String, TText> id2preprocessedText, Nicolas nicolas) throws NicolasException {
59 Map<String, String> id2summary = Maps.newHashMap(); 59 Map<String, String> id2summary = Maps.newHashMap();
  60 + int textNumber = 0;
60 for (Map.Entry<String, TText> entry : id2preprocessedText.entrySet()) { 61 for (Map.Entry<String, TText> entry : id2preprocessedText.entrySet()) {
  62 + if (textNumber++ % 10 == 0) {
  63 + LOG.info("{} texts summarized.", textNumber);
  64 + }
61 TText text = entry.getValue(); 65 TText text = entry.getValue();
62 int targetSize = calculateTargetSize(text); 66 int targetSize = calculateTargetSize(text);
63 String summary = nicolas.summarizeThrift(text, targetSize); 67 String summary = nicolas.summarizeThrift(text, targetSize);
nicolas-train/src/main/java/pl/waw/ipipan/zil/summ/nicolas/train/model/Settings.java
  View file @1b04826
@@ -4,6 +4,7 @@ import weka.classifiers.Classifier; @@ -4,6 +4,7 @@ import weka.classifiers.Classifier;
4 import weka.classifiers.meta.AttributeSelectedClassifier; 4 import weka.classifiers.meta.AttributeSelectedClassifier;
5 import weka.classifiers.trees.LMT; 5 import weka.classifiers.trees.LMT;
6 import weka.classifiers.trees.RandomForest; 6 import weka.classifiers.trees.RandomForest;
  7 +import weka.classifiers.trees.RandomTree;
7 8
8 public class Settings { 9 public class Settings {
9 10
@@ -16,6 +17,10 @@ public class Settings { @@ -16,6 +17,10 @@ public class Settings {
16 17
17 public static Classifier getMentionClassifier() { 18 public static Classifier getMentionClassifier() {
18 RandomForest classifier = new RandomForest(); 19 RandomForest classifier = new RandomForest();
  20 + RandomTree tree = new RandomTree();
  21 + tree.setMaxDepth(10);
  22 + tree.setMinNum(2);
  23 + classifier.setClassifier(tree);
19 classifier.setNumIterations(NUM_ITERATIONS); 24 classifier.setNumIterations(NUM_ITERATIONS);
20 classifier.setSeed(SEED); 25 classifier.setSeed(SEED);
21 classifier.setNumExecutionSlots(NUM_EXECUTION_SLOTS); 26 classifier.setNumExecutionSlots(NUM_EXECUTION_SLOTS);
@@ -24,6 +29,10 @@ public class Settings { @@ -24,6 +29,10 @@ public class Settings {
24 29
25 public static Classifier getSentenceClassifier() { 30 public static Classifier getSentenceClassifier() {
26 RandomForest classifier = new RandomForest(); 31 RandomForest classifier = new RandomForest();
  32 + RandomTree tree = new RandomTree();
  33 + tree.setMaxDepth(10);
  34 + tree.setMinNum(2);
  35 + classifier.setClassifier(tree);
27 classifier.setNumIterations(NUM_ITERATIONS); 36 classifier.setNumIterations(NUM_ITERATIONS);
28 classifier.setSeed(SEED); 37 classifier.setSeed(SEED);
29 classifier.setNumExecutionSlots(NUM_EXECUTION_SLOTS); 38 classifier.setNumExecutionSlots(NUM_EXECUTION_SLOTS);
nicolas-train/src/main/java/pl/waw/ipipan/zil/summ/nicolas/train/pipeline/TrainAllModels.java
  View file @1b04826
@@ -44,6 +44,7 @@ public class TrainAllModels { @@ -44,6 +44,7 @@ public class TrainAllModels {
44 LOG.info("Building classifier..."); 44 LOG.info("Building classifier...");
45 classifier.buildClassifier(instances); 45 classifier.buildClassifier(instances);
46 LOG.info("...done. Build classifier: {}", classifier); 46 LOG.info("...done. Build classifier: {}", classifier);
  47 + instances.clear();
47 48
48 String target = TARGET_MODEL_DIR + targetPath; 49 String target = TARGET_MODEL_DIR + targetPath;
49 LOG.info("Saving classifier at: {}", target); 50 LOG.info("Saving classifier at: {}", target);