private static void ReadData(ExecutionOption options, List <Article> articles)
{
if (!File.Exists(options.DataFile))
{
throw new ArgumentException(string.Format("File not found {0}", options.DataFile));
}
else
{
using (var sr = new StreamReader(options.DataFile))
{
var reader = new CsvReader(sr);
reader.Configuration.HasHeaderRecord = false;
while (reader.Read())
{
var articleId = reader.GetField <int>(0);
var wordId = reader.GetField <int>(1);
var wordCount = reader.GetField <int>(2);
articles[articleId - 1].WordCounts[wordId] = wordCount;
if (k < wordId)
{
k = wordId;
}
}
}
}
}
private static List <string> ReadGroups(ExecutionOption options)
{
List <string> groups = new List <string>();
if (!File.Exists(options.GroupsFile))
{
throw new ArgumentException(string.Format("File not found {0}", options.LabelFile));
}
else
{
using (var sr = new StreamReader(options.GroupsFile))
{
var reader = new CsvReader(sr);
reader.Configuration.HasHeaderRecord = false;
while (reader.Read())
{
var group = reader.GetField <string>(0);
groups.Add(group);
}
}
}
return(groups);
}
private static List <Article> ReadLabels(ExecutionOption options, List <string> groups)
{
List <Article> articles = new List <Article>();
if (!File.Exists(options.LabelFile))
{
throw new ArgumentException(string.Format("File not found {0}", options.LabelFile));
}
else
{
using (var sr = new StreamReader(options.LabelFile))
{
var reader = new CsvReader(sr);
reader.Configuration.HasHeaderRecord = false;
while (reader.Read())
{
int label = reader.GetField <int>(0);
Article a = new Article
{
GroupId = label - 1
};
articles.Add(a);
}
}
}
return(articles);
}
static void Main(string[] args)
{
ExecutionOption options = new ExecutionOption();
if (CommandLine.Parser.Default.ParseArguments(args, options))
{
List <string> groups = ReadGroups(options);
List <Article> articles = ReadLabels(options, groups);
ReadData(options, articles);
/*
* foreach (Article article in articles.Take(5))
* {
* Console.WriteLine(article);
* }
*/
if (options.ExecutionMode == ExecutionMode.AllWayAverage)
{
double[,] sum = new double[groups.Count, groups.Count];
int[,] count = new int[groups.Count, groups.Count];
for (int i = 0; i < articles.Count; i++)
{
Article ai = articles[i];
for (int j = i + 1; j < articles.Count; j++)
{
Article aj = articles[j];
sum[ai.GroupId, aj.GroupId] += GetDistance(ai.WordCounts, aj.WordCounts, options.SimilarityFunction);
count[ai.GroupId, aj.GroupId] += 1;
}
}
using (var sr = new StreamWriter("output.csv"))
{
var writer = new CsvWriter(sr);
for (int i = 0; i < groups.Count; i++)
{
for (int j = 0; j < groups.Count; j++)
{
double average = j > i ? sum[i, j] / count[i, j] : sum[j, i] / count[j, i];
int fieldCount = j > i ? count[i, j] : count[j, i];
if (fieldCount > 0)
{
writer.WriteField(average);
}
else
{
writer.WriteField(0.0);
}
}
writer.NextRecord();
}
}
}
else if (options.ExecutionMode == ExecutionMode.MostNearestNeighbor)
{
int[,] count = new int[groups.Count, groups.Count];
for (int i = 0; i < articles.Count; i++)
{
double maxSum = 0.0;
int maxGroupId = -1;
Article ai = articles[i];
for (int j = 0; j < articles.Count; j++)
{
Article aj = articles[j];
if (ai.GroupId == aj.GroupId)
{
continue;
}
double distance = GetDistance(ai.WordCounts, aj.WordCounts, options.SimilarityFunction);
if (distance > maxSum)
{
maxSum = distance;
maxGroupId = aj.GroupId;
}
}
if (maxGroupId >= 0)
{
count[ai.GroupId, maxGroupId] += 1;
}
}
using (var sr = new StreamWriter("output.csv"))
{
var writer = new CsvWriter(sr);
for (int i = 0; i < groups.Count; i++)
{
for (int j = 0; j < groups.Count; j++)
{
writer.WriteField(count[i, j]);
}
writer.NextRecord();
}
}
}
else if (options.ExecutionMode == ExecutionMode.MostNearestNeighborWithRandomProjection)
{
var M = Matrix <double> .Build;
var matrix = M.Random(options.RandomProjectionDimCount, k + 1, new Normal(0.0, 1.0));
DimensionReduceArticles(articles, matrix);
int[,] count = new int[groups.Count, groups.Count];
for (int i = 0; i < articles.Count; i++)
{
double maxSum = 0.0;
int maxGroupId = -1;
Article ai = articles[i];
for (int j = 0; j < articles.Count; j++)
{
Article aj = articles[j];
if (ai.GroupId == aj.GroupId)
{
continue;
}
double distance = GetDistance(ai.WordCountVector, aj.WordCountVector, options.SimilarityFunction);
if (distance > maxSum)
{
maxSum = distance;
maxGroupId = aj.GroupId;
}
}
if (maxGroupId >= 0)
{
count[ai.GroupId, maxGroupId] += 1;
}
}
using (var sr = new StreamWriter("output.csv"))
{
var writer = new CsvWriter(sr);
for (int i = 0; i < groups.Count; i++)
{
for (int j = 0; j < groups.Count; j++)
{
writer.WriteField(count[i, j]);
}
writer.NextRecord();
}
}
}
else if (options.ExecutionMode == ExecutionMode.RandomProjectionCosineScatter)
{
throw new NotSupportedException("Unknown execution mode");
}
else
{
throw new NotSupportedException("Unknown execution mode");
}
Console.WriteLine("----------------------------------------------------------------------");
}
}