public CIdentifyEntity() { CStringTool.ReadPluralToSingual(@"D:\cuiwanyun\qa\data\ResourceData\StemmingVerb.txt"); CStringTool.ReadPluralToSingual2(@"D:\cuiwanyun\qa\data\ResourceData\StemmingNoun.txt"); client = new StaticHashing(); client.InputDictionary(@"D:\cuiwanyun\qa\data\EntityCategoryDict_Satorinew.txt"); }
public void Process() { StaticHashing client = new StaticHashing(); StreamReader reader; string currentLine; client.BuildHashing(250000000, 2, 4); reader = new StreamReader(@"D:\cuiwanyun\qa\data\SatoriEntityCategoryId.txt"); while ((currentLine = reader.ReadLine()) != null) { string[] tokens = currentLine.Split('\t'); client.InsertEntity(tokens[0].ToLower(), Convert.ToInt32(tokens[1])); } reader.Close(); string[] inputPaths = new string[1]; inputPaths[0] = @"D:\cuiwanyun\qa\data\TempDict0.txt"; /* inputPaths[1] = @"D:\cuiwanyun\EntityAttributeMapping\data\TempDict1.txt"; inputPaths[2] = @"D:\cuiwanyun\EntityAttributeMapping\data\TempDict2.txt"; inputPaths[3] = @"D:\cuiwanyun\EntityAttributeMapping\data\TempDict3.txt"; inputPaths[4] = @"D:\cuiwanyun\EntityAttributeMapping\data\TempDict4.txt";*/ client.OutputDictionary(inputPaths); /* client.InputDictionary(inputPaths); reader = new StreamReader(@"D:\cuiwanyun\EntityAttributeMapping\data\QueryIE01Tail05"); StreamWriter writer=new StreamWriter(@"D:\cuiwanyun\EntityAttributeMapping\data\QueryIE01Tail05Answer.txt"); Console.WriteLine("start time: " + DateTime.Now.ToString()); int entityNumber=0; while ((currentLine = reader.ReadLine()) != null) { entityNumber++; string[] tokens = currentLine.Split('\"'); var answer = client.LongestCover(tokens[1]); writer.WriteLine(tokens[1]); foreach (var item in answer) writer.WriteLine(item.Key + "\t" + item.Value); } Console.WriteLine("end time: " + DateTime.Now.ToString()); Console.WriteLine("entity number = "+entityNumber); writer.Close();*/ }