private void SetEngine(DocumentResolver resolver)
{
if (rdoTFIDF.Checked)
{
resolver.SetEngine(DocumentResolver.EngineType.BayesTFIDF);
}
if (rdoLevenshtein.Checked)
{
resolver.SetEngine(DocumentResolver.EngineType.BayesLevenshtein);
}
}
/// <summary>
/// Get a resolver object from the cache or create a new resolver object.
/// </summary>
/// <param name="authors"></param>
/// <param name="year"></param>
/// <param name="type"></param>
/// <param name="cacheKey"></param>
/// <returns></returns>
private DocumentResolver GetResolver(string authors, string year, string type, String cacheKey)
{
DocumentResolver resolver = null;
if (MemoryCache.Default[cacheKey] != null)
{
// Use cached version
resolver = (DocumentResolver)MemoryCache.Default[cacheKey];
}
else
{
// Set up resolver
resolver = new DocumentResolver();
resolver.SetEngine(DocumentResolver.EngineType.BayesTFIDF);
switch (type)
{
case "title":
resolver.SetDictionary(GetTitleDictionary(authors != string.Empty, year != string.Empty));
break;
case "part":
default:
resolver.SetDictionary(GetSegmentDictionary(authors != string.Empty, year != string.Empty));
break;
}
MemoryCache.Default.Add(cacheKey, resolver, DateTime.Now.AddMinutes(
Convert.ToDouble(ConfigurationManager.AppSettings["ResolverCacheTime"])));
}
return(resolver);
}
private static void ResolveList(bool useWordStemmer, string outputfile, string dictionaryfile, string inputfile)
{
// Initalize instances of all of the resolvers that will be tested
DocumentResolver documentResolver = new DocumentResolver();
BayesResolverEngine bayesResolver = new BayesResolverEngine();
LevenshteinResolverEngine levenshteinResolver = new LevenshteinResolverEngine();
TFIDFResolverEngine tfidfResolver = new TFIDFResolverEngine();
// Load the dictionary into all of the resolvers
Dictionary <string, string> dictionary = LoadData(dictionaryfile);
documentResolver.SetDictionary(dictionary);
bayesResolver.Dictionary = dictionary;
levenshteinResolver.Dictionary = dictionary;
tfidfResolver.Dictionary = dictionary;
// Load the data to be resolved
Dictionary <string, string> documents = LoadData(inputfile);
// Process all 100 documents with each resolver, recording the time
// each resolver takes to complete the task.
System.IO.File.AppendAllText(outputfile,
string.Format("Processing {0} documents against a dictionary with {1} entries.\r\n\r\n",
documents.Count(),
dictionary.Count()));
documentResolver.SetEngine(DocumentResolver.EngineType.BayesTFIDF);
DateTime startTime = DateTime.Now;
foreach (KeyValuePair <string, string> document in documents)
{
documentResolver.Resolve(document.Value, useWordStemmer);
}
DateTime endTime = DateTime.Now;
System.IO.File.AppendAllText(outputfile, string.Format("Bayes/TFIDF processing complete. {0} seconds elapsed.\r\n", (endTime - startTime).TotalSeconds.ToString()));
documentResolver.SetEngine(DocumentResolver.EngineType.BayesLevenshtein);
startTime = DateTime.Now;
foreach (KeyValuePair <string, string> document in documents)
{
documentResolver.Resolve(document.Value, useWordStemmer);
}
endTime = DateTime.Now;
System.IO.File.AppendAllText(outputfile, string.Format("Bayes/Levenshtein processing complete. {0} seconds elapsed.\r\n", (endTime - startTime).TotalSeconds.ToString()));
startTime = DateTime.Now;
foreach (KeyValuePair <string, string> document in documents)
{
bayesResolver.Resolve(document.Value, useWordStemmer);
}
endTime = DateTime.Now;
System.IO.File.AppendAllText(outputfile, string.Format("Bayes processing complete. {0} seconds elapsed.\r\n", (endTime - startTime).TotalSeconds.ToString()));
startTime = DateTime.Now;
foreach (KeyValuePair <string, string> document in documents)
{
levenshteinResolver.Resolve(document.Value, useWordStemmer);
}
endTime = DateTime.Now;
System.IO.File.AppendAllText(outputfile, string.Format("Levenshtein processing complete. {0} seconds elapsed.\r\n", (endTime - startTime).TotalSeconds.ToString()));
startTime = DateTime.Now;
foreach (KeyValuePair <string, string> document in documents)
{
tfidfResolver.Resolve(document.Value, useWordStemmer);
}
endTime = DateTime.Now;
System.IO.File.AppendAllText(outputfile, string.Format("TFIDF processing complete. {0} seconds elapsed.\r\n\r\n\r\n", (endTime - startTime).TotalSeconds.ToString()));
}
private static void ResolveDocument(bool useWordStemmer, string outputfile, string dictionaryfile, string document)
{
// Initalize instances of all of the resolvers that will be tested
DocumentResolver documentResolver = new DocumentResolver();
BayesResolverEngine bayesResolver = new BayesResolverEngine();
LevenshteinResolverEngine levenshteinResolver = new LevenshteinResolverEngine();
TFIDFResolverEngine tfidfResolver = new TFIDFResolverEngine();
// Load the dictionary into all of the resolvers
Dictionary <string, string> dictionary = LoadData(dictionaryfile);
documentResolver.SetDictionary(dictionary);
bayesResolver.Dictionary = dictionary;
levenshteinResolver.Dictionary = dictionary;
tfidfResolver.Dictionary = dictionary;
// Process all 100 documents with each resolver, recording the time
// each resolver takes to complete the task.
System.IO.File.AppendAllText(outputfile,
string.Format("Processing '{0}' against a dictionary with {1} entries.\r\n",
document,
dictionary.Count()));
System.IO.File.AppendAllText(outputfile, "\r\nBayes/TFIDF\r\n");
documentResolver.SetEngine(DocumentResolver.EngineType.BayesTFIDF);
List <ResolutionResult> resolutionResults = documentResolver.Resolve(document, useWordStemmer);
foreach (ResolutionResult resolutionResult in resolutionResults)
{
System.IO.File.AppendAllText(outputfile, string.Format("{0} {1} {2}\r\n", resolutionResult.Score.ToString(), resolutionResult.Key, resolutionResult.Document));
}
System.IO.File.AppendAllText(outputfile, "\r\nBayes/Levenshtein\r\n");
documentResolver.SetEngine(DocumentResolver.EngineType.BayesLevenshtein);
resolutionResults = documentResolver.Resolve(document, useWordStemmer);
foreach (ResolutionResult resolutionResult in resolutionResults)
{
System.IO.File.AppendAllText(outputfile, string.Format("{0} {1} {2}\r\n", resolutionResult.Score.ToString(), resolutionResult.Key, resolutionResult.Document));
}
System.IO.File.AppendAllText(outputfile, "\r\nBayes\r\n");
resolutionResults = bayesResolver.Resolve(document, useWordStemmer);
foreach (ResolutionResult resolutionResult in resolutionResults)
{
System.IO.File.AppendAllText(outputfile, string.Format("{0} {1} {2}\r\n", resolutionResult.Score.ToString(), resolutionResult.Key, resolutionResult.Document));
}
System.IO.File.AppendAllText(outputfile, "\r\nLevenshtein\r\n");
resolutionResults = levenshteinResolver.Resolve(document, useWordStemmer);
foreach (ResolutionResult resolutionResult in resolutionResults)
{
System.IO.File.AppendAllText(outputfile, string.Format("{0} {1} {2}\r\n", resolutionResult.Score.ToString(), resolutionResult.Key, resolutionResult.Document));
}
System.IO.File.AppendAllText(outputfile, "\r\nTFIDF\r\n");
resolutionResults = tfidfResolver.Resolve(document, useWordStemmer);
foreach (ResolutionResult resolutionResult in resolutionResults)
{
System.IO.File.AppendAllText(outputfile, string.Format("{0} {1} {2}\r\n", resolutionResult.Score.ToString(), resolutionResult.Key, resolutionResult.Document));
}
System.IO.File.AppendAllText(outputfile, "\r\n\r\n");
}
