diff --git a/vipra-cmd/src/main/java/de/vipra/cmd/lda/Analyzer.java b/vipra-cmd/src/main/java/de/vipra/cmd/lda/Analyzer.java
index 365fc9a44d8c1cb06b4a1aa8fffa3c683fb1a3e4..f3398bb1026057f75b27406cf21491cc5a94ac7c 100644
--- a/vipra-cmd/src/main/java/de/vipra/cmd/lda/Analyzer.java
+++ b/vipra-cmd/src/main/java/de/vipra/cmd/lda/Analyzer.java
@@ -133,10 +133,10 @@ public class Analyzer {
final int topicCount = modelConfig.getkTopics();
assert topicCount > 0;
- final int sequencesCount = windowIndex.size();
- assert sequencesCount > 0;
+ final int windowCount = windowIndex.size();
+ assert windowCount > 0;
- final int articlesCount = idDateIndex.size();
+ final int articleCount = idDateIndex.size();
final int wordCount = wordIndex.size();
// read topic distributions
@@ -146,8 +146,8 @@ public class Analyzer {
throw new AnalyzerException("file not found: " + gamFile.getAbsolutePath());
in = new BufferedReader(new InputStreamReader(new FileInputStream(gamFile)));
- final double[][] topicDistributions = new double[articlesCount][topicCount];
- for (int idxArticle = 0; idxArticle < articlesCount; idxArticle++) {
+ final double[][] topicDistributions = new double[articleCount][topicCount];
+ for (int idxArticle = 0; idxArticle < articleCount; idxArticle++) {
// read distributions into matrix and sum
double topicDistributionSum = 0;
for (int idxTopic = 0; idxTopic < topicCount; idxTopic++) {
@@ -166,18 +166,23 @@ public class Analyzer {
// read topic definition files and create topics
final TopicModelFull topicModel = new TopicModelFull(modelConfig.getName(), modelConfig);
- final List<WindowFull> newWindows = new ArrayList<>(sequencesCount);
- final List<SequenceFull> newSequences = new ArrayList<>(topicCount * sequencesCount);
+ final List<WindowFull> newWindows = new ArrayList<>(windowCount);
+ final List<SequenceFull> newSequences = new ArrayList<>(topicCount * windowCount);
final List<TopicFull> newTopics = new ArrayList<>(topicCount);
+ topicModel.setWordCount(wordCount);
+ topicModel.setWindowCount(windowCount);
+ topicModel.setArticleCount(articleCount);
+ topicModel.setTopicCount(topicCount);
+
ConsoleUtils.info("vocabulary size: " + wordCount);
- ConsoleUtils.info("sequences: " + sequencesCount);
+ ConsoleUtils.info("sequences: " + windowCount);
ConsoleUtils.info("topics: " + topicCount);
final boolean seqRelativeCutoff = modelConfig.getMinRelativeProbability() > 0;
// create sequence windows
- for (int idxSeq = 0; idxSeq < sequencesCount; idxSeq++) {
+ for (int idxSeq = 0; idxSeq < windowCount; idxSeq++) {
final WindowFull newWindow = new WindowFull();
newWindow.setId(idxSeq);
newWindow.setStartDate(windowIndex.startDate(idxSeq));
@@ -197,7 +202,7 @@ public class Analyzer {
// create new topic
final TopicFull newTopic = new TopicFull();
- final List<Sequence> newTopicSequences = new ArrayList<>(sequencesCount);
+ final List<Sequence> newTopicSequences = new ArrayList<>(windowCount);
newTopic.setSequences(newTopicSequences);
newTopic.setTopicModel(new TopicModel(topicModel.getId()));
newTopics.add(newTopic);
@@ -206,9 +211,9 @@ public class Analyzer {
// read file lines into word x sequence matrix
// gather maximum likeliness per sequence and per word
- final double[][] likelinesses = new double[wordCount][sequencesCount];
+ final double[][] likelinesses = new double[wordCount][windowCount];
for (int idxWord = 0; idxWord < wordCount; idxWord++) {
- for (int idxSeq = 0; idxSeq < sequencesCount; idxSeq++) {
+ for (int idxSeq = 0; idxSeq < windowCount; idxSeq++) {
likelinesses[idxWord][idxSeq] = Double.parseDouble(in.readLine());
}
}
@@ -221,12 +226,12 @@ public class Analyzer {
// collect top words in each sequence for topic name
final Set<TopicWord> topTopicWords = new HashSet<>();
- final double[] relevances = new double[sequencesCount];
+ final double[] relevances = new double[windowCount];
double relevanceSum = 0;
double prevRelevance = 0;
// for each sequence
- for (int idxSeq = 0, sequenceOffset = 0; idxSeq < sequencesCount; idxSeq++) {
+ for (int idxSeq = 0, sequenceOffset = 0; idxSeq < windowCount; idxSeq++) {
// calculate relative cutoff probability
final double maxSeqLikeliness = maxSeqLikelinesses[idxSeq];
final double minRelativeSeqLikeliness = modelConfig.getMinRelativeProbability() * Math.abs(maxSeqLikeliness);
@@ -291,14 +296,14 @@ public class Analyzer {
newTopic.setName(TopicFull.getNameFromWords(modelConfig.getTopicAutoNamingWords(), topTopicWordsList));
// calculate average
- final double average = relevanceSum / sequencesCount;
+ final double average = relevanceSum / windowCount;
newTopic.setAvgRelevance(average);
// calculate variance
double variance = 0;
for (final double relevance : relevances)
variance += Math.pow(relevance - average, 2);
- newTopic.setVarRelevance(variance / sequencesCount);
+ newTopic.setVarRelevance(variance / windowCount);
// calculate rising/falling/rising-decay relevances
double risingRelevance = 0;
@@ -312,7 +317,7 @@ public class Analyzer {
} else {
fallingRelevance += Math.abs(relevanceDiff);
}
- risingDecayRelevance += Math.exp(-modelConfig.getRisingDecayLambda() * (sequencesCount - idxSeq2 + 1)) * relevanceDiff;
+ risingDecayRelevance += Math.exp(-modelConfig.getRisingDecayLambda() * (windowCount - idxSeq2 + 1)) * relevanceDiff;
}
newTopic.setRisingRelevance(risingRelevance);
newTopic.setFallingRelevance(fallingRelevance);
@@ -356,9 +361,9 @@ public class Analyzer {
// calculate divergences
- final List<SimilarArticle> similarArticles = new ArrayList<>(articlesCount - 1);
+ final List<SimilarArticle> similarArticles = new ArrayList<>(articleCount - 1);
- for (int idxArticle2 = 0; idxArticle2 < articlesCount; idxArticle2++) {
+ for (int idxArticle2 = 0; idxArticle2 < articleCount; idxArticle2++) {
if (idxArticle == idxArticle2)
continue;
diff --git a/vipra-ui/app/index.html b/vipra-ui/app/index.html
index c39c42bec7338c08efb1ab66f3ffd92f441efb57..d2e3ede9b09901f6e0deda67c077b0d98f6dee6e 100644
--- a/vipra-ui/app/index.html
+++ b/vipra-ui/app/index.html
@@ -95,8 +95,11 @@
<div class="modal-body">
<ul class="list-group" ng-show="topicModels.length">
<button type="button" class="list-group-item" ng-repeat="topicModel in topicModels" ng-click="changeTopicModel(topicModel)" ng-class="{active:rootModels.topicModel.id===topicModel.id}">
- <span class="badge" ng-bind="topicModel.modelConfig.kTopics"></span>
+ <span class="badge" ng-bind="topicModel.articleCount" ng-show="topicModel.articleCount" ng-attr-title="{{topicModel.articleCount + ' article(s)'}}"></span>
+ <span class="badge" ng-bind="topicModel.topicCount" ng-show="topicModel.topicCount" ng-attr-title="{{topicModel.topicCount + ' topic(s)'}}"></span>
<span ng-bind="topicModel.id"></span>
+ <br ng-show="topicModel.modelConfig.description">
+ <small ng-bind="topicModel.modelConfig.description"></small>
</button>
</ul>
<p class="text-center" ng-show="loading.any">
@@ -116,7 +119,7 @@
<pre>vipra -S some_model -I data.json</pre>
</li>
<li>
- Generate topic data:
+ Generate topic data and index:
<pre>vipra -S some_model -Mi</pre>
</li>
</ol>
diff --git a/vipra-ui/app/js/controllers.js b/vipra-ui/app/js/controllers.js
index 1a4e7a39799c413e098c4a6ef47f88626b5ac3f3..f0975308cd97d5715392b414cf6675b1492b508c 100644
--- a/vipra-ui/app/js/controllers.js
+++ b/vipra-ui/app/js/controllers.js
@@ -32,7 +32,7 @@
}
TopicModelFactory.query({
- fields: 'modelConfig'
+ fields: '_all'
}, function(data) {
$scope.topicModels = data;
}, function(err) {
diff --git a/vipra-util/src/main/java/de/vipra/util/model/TopicModelFull.java b/vipra-util/src/main/java/de/vipra/util/model/TopicModelFull.java
index ad5572ab112f3f7aa426e76a86e8f8593419d936..ead4c87e29ac95dbbce1d1c23558a248390c54a7 100644
--- a/vipra-util/src/main/java/de/vipra/util/model/TopicModelFull.java
+++ b/vipra-util/src/main/java/de/vipra/util/model/TopicModelFull.java
@@ -18,7 +18,13 @@ public class TopicModelFull implements Model<String>, Comparable<TopicModelFull>
@Id
private String id;
- private String description;
+ private int topicCount;
+
+ private int articleCount;
+
+ private int wordCount;
+
+ private int windowCount;
@Embedded
@QueryIgnore(multi = true)
@@ -45,12 +51,36 @@ public class TopicModelFull implements Model<String>, Comparable<TopicModelFull>
this.id = id;
}
- public String getDescription() {
- return description;
+ public int getTopicCount() {
+ return topicCount;
+ }
+
+ public void setTopicCount(int topicCount) {
+ this.topicCount = topicCount;
+ }
+
+ public int getArticleCount() {
+ return articleCount;
+ }
+
+ public void setArticleCount(int articleCount) {
+ this.articleCount = articleCount;
+ }
+
+ public int getWordCount() {
+ return wordCount;
+ }
+
+ public void setWordCount(int wordCount) {
+ this.wordCount = wordCount;
+ }
+
+ public int getWindowCount() {
+ return windowCount;
}
- public void setDescription(final String description) {
- this.description = description;
+ public void setWindowCount(int windowCount) {
+ this.windowCount = windowCount;
}
public TopicModelConfig getModelConfig() {
diff --git a/vipra-util/src/main/java/de/vipra/util/service/MongoService.java b/vipra-util/src/main/java/de/vipra/util/service/MongoService.java
index 99cafa807da5b397d6db5fb394ae752e4d5581da..0f881b0d018ad724fb105461bfbb03f69b6eac3f 100644
--- a/vipra-util/src/main/java/de/vipra/util/service/MongoService.java
+++ b/vipra-util/src/main/java/de/vipra/util/service/MongoService.java
@@ -102,7 +102,7 @@ public class MongoService<Type extends Model<IdType>, IdType> implements Service
} else {
query.retrievedFields(false, fields);
}
- } else if (ignoredFieldsMultiQuery.length > 0) {
+ } else if (!builder.isAllFields() && ignoredFieldsMultiQuery.length > 0) {
query.retrievedFields(false, ignoredFieldsMultiQuery);
}
} else if (ignoredFieldsMultiQuery.length > 0) {
diff --git a/vipra-util/src/main/java/de/vipra/util/service/Service.java b/vipra-util/src/main/java/de/vipra/util/service/Service.java
index 839e6989561dce0b3cf09c93b3f4e77941e8f72f..e791c8d0c76a35da12a6d87a5c230270b7f3971f 100644
--- a/vipra-util/src/main/java/de/vipra/util/service/Service.java
+++ b/vipra-util/src/main/java/de/vipra/util/service/Service.java
@@ -175,6 +175,7 @@ public interface Service<Type extends Model<IdType>, IdType, E extends Exception
private List<Tuple<String, Object>> criteria;
private String[] fields;
private boolean include;
+ private boolean allFields;
private QueryBuilder() {}
@@ -265,6 +266,13 @@ public interface Service<Type extends Model<IdType>, IdType, E extends Exception
*/
public QueryBuilder fields(final boolean include, final String... strings) {
this.include = include;
+ for (String field : strings) {
+ if (field.equalsIgnoreCase("_all")) {
+ this.allFields = true;
+ this.fields = null;
+ return this;
+ }
+ }
this.fields = strings;
return this;
}
@@ -295,6 +303,10 @@ public interface Service<Type extends Model<IdType>, IdType, E extends Exception
return fields;
}
+ public boolean isAllFields() {
+ return allFields;
+ }
+
}
}