// Write to an output file with the following format: place id|place name|list of words in the place name|
static public void WriteToFile(Deduplication deduplication, string outputFileName = "GroundTruth.txt", int maxOutputSize = 100)
{
int count = 0;
System.IO.StreamWriter file = new System.IO.StreamWriter(outputFileName);
foreach (var entry in deduplication.IsCoreWordProbability)
{
Place place = entry.Key;
var probabilities = entry.Value;
string line = Convert.ToString(place.id);
line += "|";
line += place.title;
line += "|";
int wordCount = 0;
foreach (var word in probabilities.Keys)
{
line += word;
++wordCount;
if (wordCount < probabilities.Count)
{
line += " ";
}
}
line += "|";
file.WriteLine(line);
++count;
if (count >= maxOutputSize)
{
break;
}
}
file.Close();
}
static void Main(string[] args)
{
Deduplication deduplication = new Deduplication();
var database = deduplication.Database;
string[] filePaths = Directory.GetFiles("../../../Data/");
foreach (string filePath in filePaths)
{
if (filePath.Contains("Paris"))
{
database.Load(filePath);
}
}
//database.GenerateTiles(0.05);
database.GenerateTilesByCity();
//database.AddPlace("Starbucks Coffee");
//database.AddPlace("Peets Coffee");
//database.AddPlace("Starbucks");
deduplication.Setup();
// Run the expectation maximization algorithm
deduplication.ExpectationMaximization(100, 1E-3, Deduplication.Model.Name);
//var cp = deduplication.IsCoreWordProbability[database.GetByName("Metro-Station Mouton Duvernet (Linie 4)")[0]];
Test test = new Test();
test.LoadGroundTruthFromFile("GroundTruth.txt");
System.IO.StreamWriter file = new System.IO.StreamWriter("precision SpatialContext.csv");
for (int i = 1; i < 15; ++i)
{
double precision = test.GetPrecision(deduplication.IsCoreWordProbability, i * 10);
file.WriteLine(i * 10 + "," + precision);
}
file.Close();
}
