public WordFrequency(String dir) {
	directory = dir;
	stopWords = new LinkedList<String>();
	textualContentFromTitle = new LinkedList<String>();
	textualContentFromHeadings = new LinkedList<String>();
	textualContentFromTables = null;
	textualContentFromLists = new LinkedList<String>();
	textualContentFromFont = null;
	textualContentFromBody = new LinkedList<String>();
	textualContentFromAnchors = new LinkedList<String>();
	textualContentFromThal = new LinkedList<String>();
	wordsBodyFrequenciesMap = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	wordsThalFrequenciesMap = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	dataBody = new DefaultDataset();
	dataThal = new DefaultDataset();
}
 

public UtilsDataset(String app) {
	outputDirectory = app;
	tagsFrequencyMatrix = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	wordsBodyFrequencyMatrix = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	wordsThalFrequencyMatrix = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	urlsDistancesMatrix = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	domsLevenshteinDistancesMatrix = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	domsRobustTreeEditDistancesMatrix = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	tagsFrequenciesDataset = new DefaultDataset();
	wordsBodyFrequenciesDataset = new DefaultDataset();
	wordsThalFrequenciesDataset = new DefaultDataset();
	urlsDistancesDataset = new DefaultDataset();
	domsLevenshteinDistancesDataset = new DefaultDataset();
	domsRobustTreeEditDistancesDataset = new DefaultDataset();
}
 

public DomDistance(String dir, Distance dist) {
	directory = dir;
	distance = dist;
	tagsMasterVector = new LinkedList<String>();
	domDistancesMap = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	data = new DefaultDataset();
	stringsToRemove = new LinkedList<String>();
}
 

/**
 * creates an instance of TagFrequency class
 * 
 * @param dir
 */
public TagFrequency(String dir) {
	directory = dir;
	tagsMasterVector = new LinkedList<String>();
	tagsFrequenciesMap = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
	data = new DefaultDataset();
}
 

/**
 * Access the data as a Dataset, for kmeans
 * @return a Dataset representing the data for kmeans and javaml applications
 */
public Dataset getDataSet(){
	Dataset data = new DefaultDataset();
	for (int i = 0;i < tracks.size();i++){
		DenseInstance ins = new DenseInstance(tracks.get(i));
		//for (int a = 0;a < tracks.get(i).length;a++){
			//System.out.println(tracks.get(i)[a]);
		//}
		data.add(ins);
	}
	
	return data;
}
 

public UrlDistance(String dir) {
	directory = dir;
	data = new DefaultDataset();
	urlDistancesMap = new LinkedHashMap<String, LinkedHashMap<String, BigDecimal>>();
}