Go to the documentation of this file.00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012 #include <shogun/lib/common.h>
00013 #include <shogun/distance/ManhattanWordDistance.h>
00014 #include <shogun/features/Features.h>
00015 #include <shogun/features/StringFeatures.h>
00016 #include <shogun/io/SGIO.h>
00017
00018 using namespace shogun;
00019
00020 CManhattanWordDistance::CManhattanWordDistance()
00021 : CStringDistance<uint16_t>()
00022 {
00023 SG_DEBUG("CManhattanWordDistance created");
00024 }
00025
00026 CManhattanWordDistance::CManhattanWordDistance(
00027 CStringFeatures<uint16_t>* l, CStringFeatures<uint16_t>* r)
00028 : CStringDistance<uint16_t>()
00029 {
00030 SG_DEBUG("CManhattanWordDistance created");
00031
00032 init(l, r);
00033 }
00034
00035 CManhattanWordDistance::~CManhattanWordDistance()
00036 {
00037 cleanup();
00038 }
00039
00040 bool CManhattanWordDistance::init(CFeatures* l, CFeatures* r)
00041 {
00042 bool result=CStringDistance<uint16_t>::init(l,r);
00043 return result;
00044 }
00045
00046 void CManhattanWordDistance::cleanup()
00047 {
00048 }
00049
00050 float64_t CManhattanWordDistance::compute(int32_t idx_a, int32_t idx_b)
00051 {
00052 int32_t alen, blen;
00053 bool free_avec, free_bvec;
00054
00055 uint16_t* avec=((CStringFeatures<uint16_t>*) lhs)->
00056 get_feature_vector(idx_a, alen, free_avec);
00057 uint16_t* bvec=((CStringFeatures<uint16_t>*) rhs)->
00058 get_feature_vector(idx_b, blen, free_bvec);
00059
00060 int32_t result=0;
00061
00062 int32_t left_idx=0;
00063 int32_t right_idx=0;
00064
00065 while (left_idx < alen && right_idx < blen)
00066 {
00067 uint16_t sym=avec[left_idx];
00068 if (avec[left_idx]==bvec[right_idx])
00069 {
00070 int32_t old_left_idx=left_idx;
00071 int32_t old_right_idx=right_idx;
00072
00073 while (left_idx< alen && avec[left_idx]==sym)
00074 left_idx++;
00075
00076 while (right_idx< blen && bvec[right_idx]==sym)
00077 right_idx++;
00078
00079 result += CMath::abs( (left_idx-old_left_idx) - (right_idx-old_right_idx) );
00080 }
00081 else if (avec[left_idx]<bvec[right_idx])
00082 {
00083
00084 while (left_idx< alen && avec[left_idx]==sym)
00085 {
00086 result++;
00087 left_idx++;
00088 }
00089 }
00090 else
00091 {
00092 sym=bvec[right_idx];
00093
00094 while (right_idx< blen && bvec[right_idx]==sym)
00095 {
00096 result++;
00097 right_idx++;
00098 }
00099 }
00100 }
00101
00102 result+=blen-right_idx + alen-left_idx;
00103
00104 ((CStringFeatures<uint16_t>*) lhs)->
00105 free_feature_vector(avec, idx_a, free_avec);
00106 ((CStringFeatures<uint16_t>*) rhs)->
00107 free_feature_vector(bvec, idx_b, free_bvec);
00108
00109 return result;
00110 }
00111