public void CompareStringMetrics(string s1, string s2) { var consine = new F23.StringSimilarity.Cosine().Distance(s1, s2); var damerau = new F23.StringSimilarity.Damerau().Distance(s1, s2); var jaccard = new F23.StringSimilarity.Jaccard().Distance(s1, s2); var jarodwinkler = new F23.StringSimilarity.JaroWinkler().Distance(s1, s2); var levenshtein = new F23.StringSimilarity.Levenshtein().Distance(s1, s2); var lcs = new F23.StringSimilarity.LongestCommonSubsequence().Distance(s1, s2); var mlcs = new F23.StringSimilarity.MetricLCS().Distance(s1, s2); var ng = new F23.StringSimilarity.NGram().Distance(s1, s2); var nl = new F23.StringSimilarity.NormalizedLevenshtein().Distance(s1, s2); var osa = new F23.StringSimilarity.OptimalStringAlignment().Distance(s1, s2); var qg = new F23.StringSimilarity.QGram().Distance(s1, s2); //var ng = new F23.StringSimilarity.ShingleBased var sd = new F23.StringSimilarity.SorensenDice().Distance(s1, s2); //var wl = new F23.StringSimilarity.WeightedLevenshtein().Distance Console.WriteLine("consine score = {0}", consine.ToString()); Console.WriteLine("damerau score = {0}", damerau.ToString()); Console.WriteLine("jaccard score = {0}", jaccard.ToString()); Console.WriteLine("jarodwinkler score = {0}", jarodwinkler.ToString()); Console.WriteLine("levenshtein score = {0}", levenshtein.ToString()); Console.WriteLine("lcs score = {0}", lcs.ToString()); Console.WriteLine("mlcs score = {0}", mlcs.ToString()); Console.WriteLine("ng score = {0}", ng.ToString()); Console.WriteLine("nl score = {0}", nl.ToString()); Console.WriteLine("osa score = {0}", osa.ToString()); Console.WriteLine("qg score = {0}", qg.ToString()); //Console.WriteLine("ng score = {0}",ng.ToString()); Console.WriteLine("sd score = {0}", sd.ToString()); //Console.WriteLine("wl score = {0}",wl.ToString()); }
public List <FullAddressFields> getBestAddresses(string user_county, string[] user_search_normalized) { TextFieldParser parser = new TextFieldParser(addrPath); parser.TextFieldType = FieldType.Delimited; parser.SetDelimiters(","); //Dictionary<int, string> header = new Dictionary<int, string>(); string[] fields = parser.ReadFields(); int fields_len = fields.Length; int county_index = -1; int locality_index = -1; int secLocality_index = -1; int thorofare_index = -1; int address_index = -1; int building_id_index = -1; int address_reference_index = -1; int building_number_index = -1; int building_group_name_index = -1; int building_name_index = -1; int sub_building_name_index = -1; int department_index = -1; int organisation_name_index = -1; for (int col_index = 0; col_index < fields_len; col_index++) { string value = fields.GetValueAt <string>(col_index); if (string.Equals(county, value, StringComparison.OrdinalIgnoreCase)) { county_index = col_index; } else if (string.Equals(locality, value, StringComparison.OrdinalIgnoreCase)) { locality_index = col_index; } else if (string.Equals(secLocality, value, StringComparison.OrdinalIgnoreCase)) { secLocality_index = col_index; } else if (string.Equals(thorofare, value, StringComparison.OrdinalIgnoreCase)) { thorofare_index = col_index; } else if (string.Equals(address, value, StringComparison.OrdinalIgnoreCase)) { address_index = col_index; } else if (string.Equals(building, value, StringComparison.OrdinalIgnoreCase)) { building_id_index = col_index; } else if (string.Equals(address_reference, value, StringComparison.OrdinalIgnoreCase)) { address_reference_index = col_index; } else if (string.Equals(building_number, value, StringComparison.OrdinalIgnoreCase)) { building_number_index = col_index; } else if (string.Equals(building_group_name, value, StringComparison.OrdinalIgnoreCase)) { building_group_name_index = col_index; } else if (string.Equals(building_name, value, StringComparison.OrdinalIgnoreCase)) { building_name_index = col_index; } else if (string.Equals(sub_building_name, value, StringComparison.OrdinalIgnoreCase)) { sub_building_name_index = col_index; } else if (string.Equals(department, value, StringComparison.OrdinalIgnoreCase)) { department_index = col_index; } else if (string.Equals(organisation_name, value, StringComparison.OrdinalIgnoreCase)) { organisation_name_index = col_index; } //header.Add(col_id, fields.GetValueAt<string>(col_id)); } System.Collections.Generic.HashSet <FullAddressFields> top_addresses = new System.Collections.Generic.HashSet <FullAddressFields>(); List <int> address_scores = new List <int>(); while (!parser.EndOfData) { fields = parser.ReadFields(); // build json using full address fields FullAddressFields fulladdr = new FullAddressFields(); fulladdr.locality = fields.GetValueAt <string>(locality_index); fulladdr.secondary_locality = fields.GetValueAt <string>(secLocality_index); fulladdr.thorofare = fields.GetValueAt <string>(thorofare_index); fulladdr.building_number = fields.GetValueAt <string>(building_number_index); fulladdr.county = fields.GetValueAt <string>(county_index); fulladdr.address = fields.GetValueAt <string>(address_index); fulladdr.building = fields.GetValueAt <int>(building_id_index); fulladdr.address_reference = fields.GetValueAt <long>(address_reference_index); fulladdr.building_group_name = fields.GetValueAt <string>(building_group_name_index); fulladdr.building_name = fields.GetValueAt <string>(building_name_index); fulladdr.sub_building_name = fields.GetValueAt <string>(sub_building_name_index); fulladdr.department = fields.GetValueAt <string>(department_index); fulladdr.organisation_name = fields.GetValueAt <string>(organisation_name_index); if (string.Equals(user_county, fulladdr.county, StringComparison.OrdinalIgnoreCase)) { string address_split = fulladdr.address.Replace(fulladdr.address.Split(",").Last(), ""); string[] address_split_normalized = ner.normalize(address_split); if (!fulladdr.building_number.IsEmpty()) { if (!ner.numbers.Contains <string>(fulladdr.building_number.ToLower().Trim())) { continue; } } if (!fulladdr.sub_building_name.IsEmpty()) { string sub_building_number = Regex.Replace(fulladdr.sub_building_name, @"(apartment)|(unit)|(flat)", "", RegexOptions.IgnoreCase).ToLower().Trim(); if (!ner.numbers.Contains <string>(sub_building_number)) { continue; } } int tokens_match = 0; var sim = new F23.StringSimilarity.JaroWinkler(); var dist = new F23.StringSimilarity.Damerau(); foreach (string user_search_token in user_search_normalized) { foreach (string address_token in address_split_normalized) { //double score = sim.Similarity(user_search_token, address_token); double score = Fuzz.WeightedRatio(user_search_token, address_token, PreprocessMode.Full); double score2 = dist.Distance(user_search_token, address_token); if (score >= MIN_SCORE * 100 || score2 <= MAX_DIST) //if (score >= MIN_SCORE && score2 <= MAX_DIST) { tokens_match += 1; break; } } } if (user_search_normalized.Length == tokens_match) { top_addresses.Add(fulladdr); } //break; } } List <FullAddressFields> best_addresses = new List <FullAddressFields>(); foreach (FullAddressFields addr in top_addresses) { if (!addr.building_number.IsEmpty() || !addr.sub_building_name.IsEmpty()) { best_addresses.Add(addr); } } if (best_addresses.Count == 0) { return(top_addresses.ToList <FullAddressFields>()); } return(best_addresses); }