private Double CosineSimilarity(HashMap<String, Double> table1, HashMap<String, Double> table2)
throws Exception {
if (table1.size() != table2.size()) {
throw new Exception("Table sizes must be equal");
}
// length of table 1
double length1 = 0;
double length2 = 0;
// Double firstValue;
double secValue;
// sum of vector multiplication
double svMul = 0;
for (Entry<String, Double> kv : table1.entrySet()) {
length1 += Math.pow(kv.getValue(), 2);
secValue = table2.get(kv.getKey());
length2 += Math.pow(secValue, 2);
svMul += secValue * kv.getValue();
}
length1 = Math.sqrt(length1);
length2 = Math.sqrt(length2);
return Double.parseDouble(NumericFormat.getNumberFormated(svMul / (length1 * length2)));
}
public Double GetCosineSimilarity() throws Exception {
HashMap<String, Double> frequencyTable1 = PrepareFrequency(words1);
HashMap<String, Double> frequencyTable2 = PrepareFrequency(words2);
HashMap<String, Double> tfTable1 = TfFactorized(frequencyTable1);
HashMap<String, Double> tfTable2 = TfFactorized(frequencyTable2);
LinkedList<HashMap<String, Double>> tables = new LinkedList<HashMap<String, Double>>();
tables.add(tfTable1);
tables.add(tfTable2);
PrepareAllHashTables(tables);
tables = GetPreparedTFIDFTables(IDFDocumentTable(tables), tables);
return Double.parseDouble(
NumericFormat.getNumberFormated(CosineSimilarity(tables.get(0), tables.get(1))));
}
