// This method tests if an expected match is found for a given ticket and data set
public bool ContainsMatch(ICompressible testTicket, ICompressible expectedMatch, ICompressible[] dataSet)
{
// Similarity object to use for FindSimilarEntities
Similarity simTest = new Similarity();
// Get the ordered results and return if the match is present
ICompressible[] results = simTest.FindSimilarEntities(testTicket, dataSet);
if(results.Contains(expectedMatch))
{
return true;
}
else
{
return false;
}
}
static void Main(string[] args)
{
//Timer to record search time
Stopwatch timer;
//Similarity object to use for the FindSimilarEntities function
Similarity simObject = new Similarity();
simObject.Threshold = 0.45;
//The ID of the ticket that you want to find similar tickets for
string searchID;
//List of tickets and their NCD values that match the search ticket
List<Tuple<double, StringCompressible>> results = new List<Tuple<double, StringCompressible>>();
//Create the data set as a list of StringCompressible objects
List<StringCompressible> DataSet = new List<StringCompressible>();
//Open the CSV to read in the data set
string currentDirectory = Directory.GetCurrentDirectory();
var CSVReader = new StreamReader(File.OpenRead(Path.Combine(currentDirectory, "IncidentRequest_Gold5k.csv")));
//Read in the "golden set" and add the entities to DataSet
while (!CSVReader.EndOfStream)
{
var row = CSVReader.ReadLine();
DataSet.Add(new StringCompressible(row.Substring(0, row.IndexOf(',')), row.Substring(row.IndexOf(',') + 1)));
//Console.WriteLine("itemID: {0}, summary: {1}", row.Substring(0, row.IndexOf(',')), row.Substring(row.IndexOf(',')+1));
}
while (true)
{
//Get the item ID to search for and store it in searchID
Console.WriteLine("Enter the itemID of the search ticket ('q' to quit): ");
searchID = Console.ReadLine();
//REPL control
if (searchID.Equals("q"))
{
break;
}
//Start the timer before searching
timer = Stopwatch.StartNew();
// Find the itemID matching the requested ticket and populate the results List with similar tickets
foreach (StringCompressible ticket in DataSet)
{
if (ticket.ItemID.Equals(searchID))
{
results = simObject.FindSimilarValAndEntities(ticket, DataSet.ToArray());
}
}
int counter = 1; // Counter for the number of tickets to return
// Print output formatting
Console.WriteLine();
Console.WriteLine("-----Similar Ticket List-----");
Console.WriteLine();
// Print the similar tickets with their ID and NCD/MCD value
foreach (Tuple<double, StringCompressible> ticket in results)
{
//if (counter > 20)
//{
// break;
//}
Console.WriteLine("{0}.\tTicket ID: {1}\tConfidence Rating: {2}", counter, ticket.Item2.ItemID, ticket.Item1);
counter++;
}
timer.Stop(); // Stop the timer
// Output the time it took to return the results
Console.WriteLine();
Console.WriteLine("Searched 5,000 tickets and produced results in {0} ms", timer.ElapsedMilliseconds);
Console.WriteLine();
}
}
public void TestDetection()
{
//Create the data set as a list of StringCompressible objects
List<StringCompressible> DataSet = new List<StringCompressible>();
//Open the CSV to read in the data set
string currentDirectory = Directory.GetCurrentDirectory();
var CSVReader = new StreamReader(File.OpenRead(Path.Combine(currentDirectory, "IncidentRequest_Gold.csv")));
//Read in the "golden set" and add the entities to DataSet
while (!CSVReader.EndOfStream)
{
var row = CSVReader.ReadLine();
var data = row.Split(',');
DataSet.Add(new StringCompressible(data[0], data[1]));
}
//Open log file for writing
var logFile = new StreamWriter(Path.Combine(currentDirectory, "TestDetection_Log.txt"));
//Create the expected outcome 2D list
List<List<string>> expectedLists = new List<List<string>>();
//Read the expected outcomes json file into a string
StreamReader fileReader = new StreamReader(File.OpenRead(Path.Combine(currentDirectory, "expectedOutcomes.json")));
string jsonText = fileReader.ReadToEnd();
JsonTextReader JReader = new JsonTextReader(new StringReader(jsonText));
//Populate the expectedLists 2D list with the expected outcomes
while (JReader.Read())
{
List<string> expectedMatches = new List<string>();
if (JReader.TokenType.ToString() == "PropertyName" && JReader.Value.ToString() != "expectedOutcomes")
{
expectedMatches.Add(JReader.Value.ToString());
JReader.Read();
JReader.Read();
while (JReader.TokenType.ToString() != "EndArray")
{
expectedMatches.Add(JReader.Value.ToString());
JReader.Read();
}
expectedLists.Add(expectedMatches);
}
}
//Similarity object to use for FindSimilarEntities
Similarity simTest = new Similarity();
int currentList = 0;
//Get the ordered results and return if the match is present
foreach(StringCompressible ticket in DataSet)
{
ICompressible[] results = simTest.FindSimilarEntities(ticket, DataSet.ToArray());
logFile.Write("{0} Matches: ", expectedLists[currentList][0]);
foreach (StringCompressible expectedMatch in results)
{
logFile.Write("{0} ", expectedMatch.ItemID);
logFile.Write("({0}), ", expectedLists[currentList].Contains(expectedMatch.ItemID));
}
currentList++;
logFile.WriteLine();
}
logFile.Close();
//Read the log file and assert that there are no false positives
var logReader = new StreamReader(File.OpenRead(Path.Combine(currentDirectory, "TestDetection_Log.txt")));
string log = logReader.ReadToEnd();
logReader.Close();
Assert.IsFalse(log.Contains("False"));
}
public void TestSymmetry()
{
byte[] xData = Encoding.ASCII.GetBytes("Lorem ipsum dolor sit amet,");
byte[] yData = Encoding.ASCII.GetBytes(" consectetur adipiscing elit.");
ICompressible forward = new MockEntity(xData.Concat(yData).ToArray());
ICompressible backward = new MockEntity(yData.Concat(xData).ToArray());
ISimilarity simTest = new Similarity();
Assert.IsTrue(simTest.GetComplexity(forward) == simTest.GetComplexity(backward));
}
public void TestMonotonicity()
{
byte[] xData = Encoding.ASCII.GetBytes("Lorem Ipsum Dolor");
byte[] appendingData = xData.Concat(Encoding.ASCII.GetBytes("sit amet, consectetur adipiscing elit")).ToArray();
ICompressible xy = new MockEntity(appendingData);
ICompressible x = new MockEntity(xData);
ISimilarity simTest = new Similarity();
Assert.IsTrue(simTest.GetComplexity(xy) >= simTest.GetComplexity(x));
}
public void TestIdempotency()
{
ICompressible x = new MockEntity(Encoding.ASCII.GetBytes("Lorem Ipsum Dolor"));
byte[] doubleArray = x.ToByteArray().Concat(x.ToByteArray()).ToArray();
byte[] emptyArray = new byte[0];
ICompressible xx = new MockEntity(doubleArray);
ICompressible e = new MockEntity(emptyArray);
ISimilarity simTest = new Similarity();
Assert.IsTrue(simTest.GetComplexity(x) == simTest.GetComplexity(xx));
Assert.IsTrue(simTest.GetComplexity(e) == 0);
}
public void TestDistributivity()
{
byte[] xData = Encoding.ASCII.GetBytes("Lorem ipsum dolor sit amet,");
byte[] yData = Encoding.ASCII.GetBytes(" consectetur adipiscing elit.");
byte[] zData = Encoding.ASCII.GetBytes("Suspendisse porttitor lectus");
ICompressible xy = new MockEntity(xData.Concat(yData).ToArray());
ICompressible xz = new MockEntity(xData.Concat(zData).ToArray());
ICompressible z = new MockEntity(zData);
ICompressible yz = new MockEntity(yData.Concat(zData).ToArray());
ISimilarity simTest = new Similarity();
Assert.IsTrue(simTest.GetComplexity(xy) + simTest.GetComplexity(z) <= simTest.GetComplexity(xz) + simTest.GetComplexity(yz));
}
