private void ProcessLinearDistanceMeasure(int docID, List <int> searchPhraseIndex)
{
string sql = "SELECT PhraseID, WordIndex FROM PhraseLocation WHERE DocumentID = " + docID + ";";
SQLiteDataReader reader = dataBase.PerformQuery(sql);
while (reader.Read())
{
int phraseID = (int)reader["PhraseID"];
int wordIndex = (int)reader["WordIndex"];
//if this isn't already in the dict add it
if (!frequencyDictionary.ContainsKey(phraseID))
{
FreqDataSet freqDataSet = new FreqDataSet();
frequencyDictionary.Add(phraseID, freqDataSet);
}
//find the min linear distance and update frequency
int minDist = int.MaxValue;
foreach (int searchIndex in searchPhraseIndex)
{
minDist = Math.Min(minDist, Math.Abs(wordIndex - searchIndex));
}
//update the changed values
frequencyDictionary[phraseID].SumLinearDistance += 1.0 / Math.Sqrt(minDist);
frequencyDictionary[phraseID].StandardFrequency++;
}
reader.Close();
}
private void CreateFrequency()
{
frequency = new List <ProximityWordFreq>();
foreach (KeyValuePair <int, FreqDataSet> entry in frequencyDictionary)
{
//check if this is a valid word
String phrase = GetPhrase(entry.Key).Phrase;
if (phrase.Length <= 1 || phrase.Equals("rrb") || phrase.Equals("lrb") || phrase.Equals(searchPhrase.Phrase))
{
continue;
}
FreqDataSet freqDataSet = entry.Value;
//add the data to the list
double linearDistance = ((int)(10000.0 * freqDataSet.SumLinearDistance / freqDataSet.StandardFrequency) / 100.0);
ProximityWordFreq proximityWordFreq = new ProximityWordFreq(phrase,
freqDataSet.StandardFrequency,
freqDataSet.NumMatchingParagraphs,
freqDataSet.NumMatchingSentences,
linearDistance);
frequency.Add(proximityWordFreq);
}
}
private void ProcessParagraphAndSentence(int docID, int paraIndex, HashSet <int> uniqueSentenceIndex)
{
//get the sentence data for the paragraph
String sql = "SELECT PhraseID, SentenceIndex FROM PhraseLocation WHERE DocumentID = " + docID + " AND ParagraphIndex = " + paraIndex + ";";
SQLiteDataReader reader = dataBase.PerformQuery(sql);
while (reader.Read())
{
int phraseID = (int)reader["phraseID"];
int sentenceIndex = (int)reader["SentenceIndex"];
// Don't process the search word when found
if (phraseID == searchPhrase.PhraseID)
{
continue;
}
//if this isn't already in the dict add it
if (!frequencyDictionary.ContainsKey(phraseID))
{
FreqDataSet freqDataSet = new FreqDataSet();
frequencyDictionary.Add(phraseID, freqDataSet);
}
frequencyDictionary[phraseID].NumMatchingParagraphs++;
//if the sentance has the target
if (uniqueSentenceIndex.Contains(sentenceIndex))
{
frequencyDictionary[phraseID].NumMatchingSentences++;
}
}
reader.Close();
}
