Class: Linkage::Comparators::Strcompare
- Inherits:
-
Linkage::Comparator
- Object
- Linkage::Comparator
- Linkage::Comparators::Strcompare
- Defined in:
- lib/linkage/comparators/strcompare.rb
Overview
Strcompare is a string comparison comparator. It uses the specified operation to compare string-type fields. Score ranges from 0 to 1.
To use Strcompare, you must specify one field for each record to use in the comparison, along with an operator. Valid operators are:
:jarowinkler
(Jaro-Winkler distance):damerau_levenshtein
(Damerau-Levenshtein distance)
Consider the following example, using a Linkage::Configuration as part of Dataset#link_with:
config.strcompare(:foo, :bar, :jarowinkler)
For each record, the values of the foo
and bar
fields are compared
using the Jaro-Winkler distance algorithm.
Damerau-Levenshtein is a modified Levenshtein that allows for transpositions It has additionally been modified to make costs of additions or deletions only 0.5
Constant Summary collapse
- VALID_OPERATIONS =
[:jarowinkler, :reverse_jarowinkler, :damerau_levenshtein]
Instance Attribute Summary
Attributes inherited from Linkage::Comparator
Instance Method Summary collapse
- #damerau_levenshtein(w1, w2) ⇒ Object
-
#initialize(field_1, field_2, operation) ⇒ Strcompare
constructor
A new instance of Strcompare.
- #jarowinkler(w1, w2) ⇒ Object
- #reverse_jarowinkler(w1, w2) ⇒ Object
- #score(record_1, record_2) ⇒ Object
Methods inherited from Linkage::Comparator
klass_for, register, #score_and_notify, #score_dataset, #score_datasets, #type, #weigh
Constructor Details
#initialize(field_1, field_2, operation) ⇒ Strcompare
Returns a new instance of Strcompare.
27 28 29 30 31 32 33 34 35 36 37 38 |
# File 'lib/linkage/comparators/strcompare.rb', line 27 def initialize(field_1, field_2, operation) if field_1.ruby_type[:type] != String || field_2.ruby_type[:type] != String raise "fields must be string types" end if !VALID_OPERATIONS.include?(operation) raise "#{operation.inspect} is not a valid operation" end @name_1 = field_1.name @name_2 = field_2.name @operation = operation end |
Instance Method Details
#damerau_levenshtein(w1, w2) ⇒ Object
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 |
# File 'lib/linkage/comparators/strcompare.rb', line 102 def damerau_levenshtein(w1, w2) a = w1.downcase b = w2.downcase aa = a.split('') ba = b.split('') al = a.length bl = b.length denom = [al, bl].max return 0 if denom == 0 oneago = nil thisrow = (1..bl).to_a + [0] al.times do |x| twoago, oneago, thisrow = oneago, thisrow, [0] * bl + [x + 1] bl.times do |y| if aa[x] == ba[y] thisrow[y] = oneago[y - 1] else delcost = oneago[y] + 0.5 addcost = thisrow[y - 1] + 0.5 subcost = oneago[y - 1] + 1 thisrow[y] = [delcost, addcost, subcost].min # remove this statement for original levenshtein if x > 0 and y > 0 and aa[x] == ba[y-1] and aa[x-1] == ba[y] thisrow[y] = [thisrow[y], twoago[y-2] + 1].min end end end end return (1 - thisrow[bl - 1] / denom.to_f).round(3) end |
#jarowinkler(w1, w2) ⇒ Object
54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 |
# File 'lib/linkage/comparators/strcompare.rb', line 54 def jarowinkler(w1, w2) a = w1.downcase b = w2.downcase aa = a.split('') ba = b.split('') al = a.length bl = b.length return 0 if al == 0 || bl == 0 l = 0 for i in Range.new(0, [[al, bl].min, 4].min-1) break if aa[i] != ba[i] l += 1 end md = [[al, bl].max/2 - 1, 1].max usea = [] useb = [] # simplify to matching characters for i in Range.new(0, al-1) fi = [[i - md, 0].max, bl-1].min li = [i + md, bl-1].min for j in Range.new(fi, li) if aa[i] == ba[j] and not useb.include?(j) usea << i useb << j break end end end bada = Range.new(0, al-1).to_a - usea badb = Range.new(0, bl-1).to_a - useb bada.reverse.each { |x| aa.delete_at(x) } badb.reverse.each { |x| ba.delete_at(x) } nm = aa.length return 0 if nm == 0 # count transpositions nt = 0 for i in Range.new(0, nm-1) nt +=1 if aa[i] != ba[i] end d = (nm/al.to_f + nm/bl.to_f + (nm-nt/2.0)/nm.to_f)/3.0 w = (d + l * 0.1 * (1 - d)).round(3) w end |
#reverse_jarowinkler(w1, w2) ⇒ Object
98 99 100 |
# File 'lib/linkage/comparators/strcompare.rb', line 98 def reverse_jarowinkler(w1, w2) jarowinkler(w1.reverse, w2.reverse) end |
#score(record_1, record_2) ⇒ Object
40 41 42 43 44 45 46 47 48 49 50 51 52 |
# File 'lib/linkage/comparators/strcompare.rb', line 40 def score(record_1, record_2) result = case @operation when :jarowinkler jarowinkler(record_1[@name_1], record_2[@name_2]) when :reverse_jarowinkler reverse_jarowinkler(record_1[@name_1], record_2[@name_2]) when :damerau_levenshtein damerau_levenshtein(record_1[@name_1], record_2[@name_2]) end result end |