public SoftTfIdf(TokenStatistics tokenStatistics, ITokenDistance distance, double tokenMatchThreshold, bool alwaysCompareLongerToShorter)
{
_tokenStatistics = tokenStatistics;
_tokenMatchThreshold = tokenMatchThreshold;
_alwaysCompareLongerToShorter = alwaysCompareLongerToShorter;
_distance = distance;
}
public SoftTfIdf(TokenStatistics tokenStatistics, ITokenDistance distance, double tokenMatchThreshold)
: this(tokenStatistics, distance, tokenMatchThreshold, false)
{
}
public SoftTfIdf(TokenStatistics tokenStatistics, ITokenDistance distance, bool alwaysCompareLongerToShorter)
: this(tokenStatistics, distance, DefaultTokenMatchThreshold, alwaysCompareLongerToShorter)
{
}
public SoftTfIdf(TokenStatistics tokenStatistics, ITokenDistance distance)
: this(tokenStatistics, distance, DefaultTokenMatchThreshold, false)
{
}
/// <summary>
/// Calculates the unit vector
/// </summary>
/// <param name="tokenStatistics"></param>
private void CalculateTfIdfs(TokenStatistics tokenStatistics)
{
double normalizer = 0.0;
double numDocuments = tokenStatistics.NumberOfDucoments;
if (numDocuments > 0)
{
foreach (var token in this.ToList())
{
int df = tokenStatistics.GetDocumentFrequency(token) ?? 1;
double w = Math.Log(GetWeight(token) + 1)*Math.Log(numDocuments/df);
SetWeight(token, w);
normalizer += w*w;
}
}
else
{
foreach (var token in this.ToList())
{
SetWeight(token, 1.0);
normalizer += 1.0;
}
}
normalizer = Math.Sqrt(normalizer);
foreach (var token in this.ToList())
{
SetWeight(token, GetWeight(token) / normalizer);
}
}
public TokenUnitVector(IEnumerable<Token> tokens, TokenStatistics tokenStatistics)
: base(tokens)
{
CalculateTfIdfs(tokenStatistics);
}
static void Main(string[] args)
{
if (args == null || args.Length == 0)
{
Console.WriteLine("Please specify the file to process. It needs to be a text file containing one name per line.");
string exeFileName = Path.GetFileName(System.Reflection.Assembly.GetEntryAssembly().Location);
Console.WriteLine(exeFileName + " [filename] [threshold 0.0-1.0]");
return;
}
string namesFilePath = args[0];
double threshold = 0.9;
if (args.Length > 1)
{
double.TryParse(args[1], out threshold);
}
try
{
var watch = Stopwatch.StartNew();
var tokenizer = new SimpleTokenizer();
var names = File.ReadAllLines(namesFilePath);
var tokenizedNameProvider = new TokenizedNameProvider(
tokenizer,
new List<ITokenTransformer>{new AbbreviationTransformer()},
new List<IStringNameOptimizer> { new DomainnameOptimizer()});
var namesTokens = tokenizedNameProvider.GetTokenizedNames(names);
var statistics = new TokenStatistics();
statistics.AddDocuments(namesTokens);
var sim = new SoftTfIdf(statistics, new JaroWinklerDistance(), 0.93, true);
var dupFinder = new TokenBasedDuplicateFinder(sim);
Console.WriteLine("Processing " + names.Length + " names.");
Console.WriteLine("Similarity Algorithm: " + sim);
Console.WriteLine("Similarity Threshold: " + threshold);
Console.WriteLine("...");
var list = dupFinder.Find(namesTokens, threshold);
var multiple = list.Where(l => l.Count > 1).ToList();
Console.WriteLine("Found " + (multiple.Sum(m => m.Count) - multiple.Count) + " duplicates.");
string resultFilePath = namesFilePath + ".result.txt";
string resultDupsOnlyFilePath = namesFilePath + ".result.dups.txt";
Console.WriteLine("Outputing " + resultFilePath);
WriteNameSetsToFile(resultFilePath, list);
Console.WriteLine("Outputing " + resultDupsOnlyFilePath);
WriteNameSetsToFile(resultDupsOnlyFilePath, multiple);
watch.Stop();
Console.WriteLine("Execution time: " + watch.ElapsedMilliseconds + "ms");
}
catch (Exception ex)
{
Console.WriteLine("Something went wrong: " + ex.Message);
}
Console.ReadKey();
}
