public static List<MasterParticipant> ReturnNonAtopicList(string filePath)
{
string[] nonAtopics = System.IO.File.ReadAllLines(filePath + "nonatopics.csv").ToArray();
List<MasterParticipant> nonADList = new List<MasterParticipant>();
foreach (string s in nonAtopics)
{
MasterParticipant participant = new MasterParticipant();
string [] splitList = s.Split(',');
string studyID = splitList[0];
participant.studyID = studyID;
nonADList.Add(participant);
}
return nonADList;
}
/// <summary>
/// Creates a list, filtering out words that are longer by 3 characters. Based on Levenshtein Distance with a poor match being > 3 edits.
/// </summary>
/// <param name="masterParticipant"></param>
/// <param name="compareTo"></param>
/// <returns></returns>
public static FirstComparison FilterByLength(MasterParticipant masterParticipant, List<ReportParticipant> compareTo)
{
int levenshteinEdits = 0;
List<ReportParticipant> levenshteinList = new List<ReportParticipant>();
foreach (ReportParticipant reportParticipant in compareTo)
{
levenshteinEdits = Math.Abs(reportParticipant.name.Length - masterParticipant.name.Length);
if (levenshteinEdits < 4 && (reportParticipant.name[0] == masterParticipant.name[0]))
{
levenshteinList.Add(reportParticipant);
}
}
return new FirstComparison
{
MasterParticipant = masterParticipant,
FilteredMatchesByLength = levenshteinList
};
}
public static string[] createStringArray(MasterParticipant masterListParticipant)
{
string[] biGrams = BiGrams.CreateNGrams(masterListParticipant.name, 2);
return biGrams;
}
