public static List <string> SentenceRankingForComparativeSum(List <string> docs, List <int> offsets, int sumLen, int lineLenght, int wordCount, int ngramsNum)
{
/*
* Sentence Ranking and extraction
* for a document represented as a fileName:
* 1) first the sentence similarity graph is formed
* 2) than the fast node (sentence) ranking is applied
* 3) top scoring sentences are extracted into final summary
*/
//Stopwatch sw = new Stopwatch();
//sw.Start();
SentenceSimilarityGraph ssGraph = new SentenceSimilarityGraph();
ssGraph.ngramsNum = ngramsNum;
ssGraph.GenerateSSGraphForComparativeSum(docs, offsets);
int nn = ssGraph.senteceNames.Count;
float beta = 0.15f;
float[] pr = new float[nn];
float[] newpr = new float[nn];
for (int i = 0; i < nn; i++)
{
pr[i] = 1 / (float)nn;
}
for (int i = 0; i < nn; i++)
{
newpr[i] = (1 - beta) / (float)nn;
}
for (int k = 0; k < 100; k++)
{
for (int i = 0; i < ssGraph.OFF.Count - 1; i++)
{
int outd = ssGraph.OFF[i + 1] - ssGraph.OFF[i];
for (int j = ssGraph.OFF[i]; j <= (ssGraph.OFF[i + 1] - 1); j++)
{
newpr[ssGraph.NB[j]] += beta * (pr[i] / outd) * ssGraph.SIGN[j];
}
}
pr = newpr;
}
Dictionary <string, float> rankedOut = new Dictionary <string, float>();
for (int k = 0; k < pr.Length; k++)
{
rankedOut[ssGraph.senteceNames[k]] = pr[k];
}
rankedOut = rankedOut.OrderByDescending(x => x.Value).ToDictionary(x => x.Key, x => x.Value);
List <string> summary = new List <string>();
int len = 0;
foreach (string line in rankedOut.Keys)
{
if (len < sumLen)
{
if (line.Length > lineLenght && Helpers.GetWords(line).Count() > wordCount)
{
int index1 = line.IndexOf("- ");
string line1;
if (index1 > 0)
{
line1 = line.Substring(index1 + 2);
}
else
{
line1 = line;
}
summary.Add(line1);
len += Helpers.GetWords(line1).Count();
}
}
}
//sw.Stop();
//Console.WriteLine(sw.ElapsedMilliseconds / (double)1000);
return(summary);
}
