//--------------------------------------------------------------------------------------------------------------------------------//
//Step-2
//1.Get the interesected profile
//2.Aplly criterion G.belongsTo(p1 and p2) if member(g,C)<n2,
// where C is a list of the most 6 common words, member() is the number of member in g that belong to C
//3.Return the intersection of the 2 profiles with only the n-gramms that satisfy criterion (2)
public DocumentProfileStopNWords ApplyMatchCriterion(DocumentProfileStopNWords profile)
{
String[] mostCommon6 = new String[] { "the", "of", "and", "a", "in", "to" };
List <String[]> nGramCollection = new List <string[]>();
//iterate through each n-gram
foreach (var ngram in profile.getNgramsCollection())
{
int membersofC = 0;
//iterate through each word in the n-gram
foreach (var word in ngram)
{
//iterate through each word in C
for (int i = 0; i < mostCommon6.Length; i++)
{
//if match is found increase members
if (word.Equals(mostCommon6[i]))
{
membersofC++;
break;
}
}
}
//if criterion (2) is satisfied add this ngram to the collection
if ((membersofC < ngram.Length))
{
nGramCollection.Add(ngram);
}
}
return(new DocumentProfileStopNWords(nGramCollection));
}
//--------------------------------------------------------------------------------------------------------------------------------//
//STEP-BY-STEP
public List <int[]> MatchedNgramSet(DocumentProfileStopNWords profile1, DocumentProfileStopNWords profile2, DocumentProfileStopNWords commonProfile)
{
List <int[]> setOfMatched = new List <int[]>();
foreach (var ngram in commonProfile.getNgramsCollection())
{
int index1 = profile1.getNgramsCollection().IndexOf(ngram);
//!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!//
//.....................IndexOF() does not work for the second profile..................//
int location = 0;
for (int index2 = 0; index2 < profile2.getNgramsCollection().Count; index2++)
{
int matches = 0;
for (int i = 0; i < profile2.getNgramsCollection()[index2].Length; i++)
{
if (profile2.getNgramsCollection()[index2][i].Equals(ngram[i]))
{
matches++;
}
}
if (matches == profile2.getNgramsCollection()[index2].Length)
{
location = index2;
}
}
setOfMatched.Add(new int[] { location, index1 });
}
return(setOfMatched);
}
//Intersect two profiles (either stopNwords or letter) and return a new intersected profile
public DocumentProfileStopNWords IntersectProfiles(DocumentProfileStopNWords profile1, DocumentProfileStopNWords profile2)
{
List <String[]> intersection = new List <string[]>();
foreach (var ngram1 in profile1.getNgramsCollection())
{
int countEquals = 0;
foreach (var ngram2 in profile2.getNgramsCollection())
{
int countEqualWords = 0;
for (int i = 0; i < ngram1.Length; i++)
{
if (ngram1[i].Equals(ngram2[i]))
{
countEqualWords++;
}
}
if (countEqualWords == ngram1.Length)
{
countEquals++;
}
}
if (countEquals > 0)
{
intersection.Add(ngram1);
}
}
DocumentProfileStopNWords profile = new DocumentProfileStopNWords(intersection);
return(profile);
}
//Step-2
//1.Get the interesected profile
//2.Aplly criterion G.belongsTo(p1 and p2) if member(g,C)<n-1 and maxseq(g,C)<n-2,
// where C is a list of the most 6 common words, member() is the number of member in g that belong to C
// and maxseq() is the maximal sequence of members of g that belong to C.
//3.Return the intersection of the 2 profiles with only the n-gramms that satisfy criterion (1)
public DocumentProfileStopNWords ApplyCanditateRetrievalCriterion(DocumentProfileStopNWords profile)
{
String[] mostCommon6 = new String[] { "the", "of", "and", "a", "in", "to" };
List <String[]> nGramCollection = new List <string[]>();
//iterate through all ngrams that belongs to the interesection of the two profiles
foreach (var ngram in profile.getNgramsCollection())
{
int maxseq = 0;
int membersofC = 0;
int currentsq = 0;
//iterate through all members of the ngram
foreach (var word in ngram)
{
bool found = false;
//iterate through all words in C
for (int i = 0; i < mostCommon6.Length; i++)
{
//increase sequence and member if match is found and stop iterating C
if (word.Equals(mostCommon6[i]))
{
membersofC++;
currentsq++;
found = true;
break;
}
}
//if match is not found compare this sequence against previous maximal and update accordingly
if (!found)
{
if (maxseq < currentsq)
{
maxseq = currentsq;
}
currentsq = 0;
}
}
//if criterion (1) is satisfied add this ngram to the collection
if ((membersofC < ngram.Length - 1) && (maxseq < ngram.Length - 2))
{
nGramCollection.Add(ngram);
}
}
return(new DocumentProfileStopNWords(nGramCollection));
}
//--------------------------------------------------------------------------------------------------------------------------------//
//STEP-BY-STEP
public DocumentProfileStopNWords GetDocumentProfileStopNWords(List <String> swPresentation, int nGramSize)
{
//calculate the size of nGram presentation
int targetIndex = swPresentation.Count + 1 - nGramSize;
List <String[]> docProfile = new List <String[]>();
//iterate through each n-gram
for (int i = 0; i < targetIndex; i++)
{
String[] ngram = new String[nGramSize];
//add words to each n-gram
for (int j = 0; j < nGramSize; j++)
{
ngram[j] = swPresentation[i + j];
}
//add current n-gram to the profile
docProfile.Add(ngram);
}
DocumentProfileStopNWords profile = new DocumentProfileStopNWords(docProfile);
return(profile);
}
//Step-1
//1.Get the profiles of the suspicious and original document
//2.Apply criterion(2)
//3.Create a list M(dx,ds) with the indexes of the common n-grams (produced in 2.) between the two profiles
// i.e. M={(1,1), (2,2), (3,5)} (ngram 1 of the plagiarized document is matched with ngram 1 of original...)
//4.Return M
public List <int[]> MatchedNgramSet(string pathToDocument1, string pathToDocument2)
{
DocumentProfileBuilder builder = new DocumentProfileBuilder();
DocumentProfileStopNWords profile1 = builder.GetDocumentProfileStopNWords(pathToDocument1, "Document", 8);
DocumentProfileStopNWords profile2 = builder.GetDocumentProfileStopNWords(pathToDocument2, "Document", 8);
Library library = new Library();
DocumentProfileStopNWords commonProfile = ApplyMatchCriterion(library.IntersectProfiles(profile1, profile2));
List <int[]> setOfMatched = new List <int[]>();
foreach (var ngram in commonProfile.getNgramsCollection())
{
int index1 = profile1.getNgramsCollection().IndexOf(ngram);
//!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!//
//.....................IndexOF() does not work for the second profile..................//
int location = 0;
for (int index2 = 0; index2 < profile2.getNgramsCollection().Count; index2++)
{
int matches = 0;
for (int i = 0; i < profile2.getNgramsCollection()[index2].Length; i++)
{
if (profile2.getNgramsCollection()[index2][i].Equals(ngram[i]))
{
matches++;
}
}
if (matches == profile2.getNgramsCollection()[index2].Length)
{
location = index2;
}
}
setOfMatched.Add(new int[] { location, index1 });
}
return(setOfMatched);
}
//Step-1
//1.Intersect the two sets which represent the two profiles
//2.Apply criterion (1)
//3.Get the suspicious n-grams
public DocumentProfileStopNWords Retrieve(DocumentProfileStopNWords profile1, DocumentProfileStopNWords profile2)
{
Library library = new Library();
return(ApplyCanditateRetrievalCriterion(library.IntersectProfiles(profile1, profile2)));
}