/**
* Returns the five most common terms of the tweetsOfEvent that belong to this event.
*
* @return A List of Strings with the most common terms.
* @see generateCommonTerms generateCommonTerms() method.
*/
public final List<String> getCommonTerms() {
if (!commonTerms.isEmpty()) {
return commonTerms;
} else {
PrintUtilities.printWarningMessageln("No common terms have been calculated yet!");
PrintUtilities.printInfoMessageln(
"Run " + PeakFindingEvent.class + "." + "generateCommonTerms() method first.");
return null;
}
}
/**
* Returns the five most common terms as a single String.
*
* @return A String containing the five most common terms.
* @see #getCommonTerms() getCommonTerms() method.
*/
public final String getCommonTermsAsString() {
if (commonTerms.isEmpty()) {
PrintUtilities.printWarningMessageln("No common terms have been calculated yet!");
PrintUtilities.printInfoMessageln(
"Run " + PeakFindingEvent.class + "." + "generateCommonTerms() method first.");
return null;
}
String commonTermsString = "";
for (String term : commonTerms) {
commonTermsString = commonTermsString + term + " ";
}
return commonTermsString;
}
/**
* Method to retrieve and store historical tweets by collecting them with their ID.
*
* @param tweetIDs The IDs of the tweets that are going to be collected.
* @param mongoDB A MongoHandler object.
* @param config A configuration object.
* @param event The ground truth event for which the tweets that are going to be collected, are
* referring to.
*/
public final void retrieveTweetsById(
List<String> tweetIDs, MongoHandler mongoDB, Config config, String event) {
ConfigurationBuilder cb = getAuthorization();
Twitter twitter = new TwitterFactory(cb.build()).getInstance();
tweetIDs
.stream()
.forEach(
(item) -> {
try {
// Get tweet and all its metadata and store it
Status status = twitter.showStatus(Long.parseLong(item));
mongoDB.insertSingleTweetIntoMongoDB(status, event);
} catch (TwitterException e) {
PrintUtilities.printErrorMessageln("Failed to retrieve tweet with ID: " + item);
Logger.getLogger(TweetsRetriever.class.getName()).log(Level.SEVERE, null, e);
}
});
}
@Override
public void apply() {
long startTime = System.currentTimeMillis();
double minTermOccur = minTermSupport * countCorpus; // Min support * Message count corpus
double maxTermOccur = maxTermSupport * countCorpus; // Max support * Message count corpus
int windows = (timeSliceB - timeSliceA) / delta2;
termDocMap = new HashMap<>();
eventList = new LinkedList<>();
PrintUtilities.printInfoMessageln("Calculating term frequencies...");
List<String> terms = corpus.getTerms();
for (int i = 0; i < terms.size(); i++) {
String term = terms.get(i);
if (term.length()
> 1) { // Stopwords check removed as they are already ommitted when creating the dataset
Integer[] frequency = corpus.getDocumentsTermFrequency(i);
int cf = 0;
for (int freq : frequency) {
cf += freq;
}
if (cf > minTermOccur && cf < maxTermOccur) {
termDocMap.put(term, frequency);
}
}
}
PrintUtilities.printInfoMessageln("Calculating windows...");
for (int i = 0; i < windows; i++) {
PrintUtilities.printInfoMessageln("Calculating window " + (i + 1) + "\n");
try {
processWindow(i);
} catch (Exception ex) {
Logger.getLogger(EDCoW.class.getName()).log(Level.SEVERE, null, ex);
}
}
Collections.sort(eventList);
events = new EDCoWEvents();
eventList
.stream()
.forEach(
(event) -> {
// try {
events.list.add(
new EDCoWEvent(
event.getKeywordsIDsAsString(),
corpus.getDateFromTimeSlice((int) event.startSlice)
+ ","
+ corpus.getDateFromTimeSlice((int) event.endSlice - 1),
corpus.getIDsOfWindowAsString(
corpus.getDateFromTimeSlice((int) event.startSlice),
corpus.getDateFromTimeSlice((int) event.endSlice - 1))));
});
events.setFullList();
long endTime = System.currentTimeMillis();
executionTime = (endTime - startTime) / 1000;
PrintUtilities.printExecutionTime(
startTime,
endTime,
EDCoW.class.getName(),
Thread.currentThread().getStackTrace()[1].getMethodName());
}