strgrp: Use angular similarity for distance metric properties

[ccan] / ccan / strgrp / strgrp.c

diff --git a/ccan/strgrp/strgrp.c b/ccan/strgrp/strgrp.c
index 00e0bf7152c32e73dc99e5fb454f224fae61852f..12119a1f276c7e4c4a9776cf061d1b8dfc77eddf 100644 (file)
--- a/ccan/strgrp/strgrp.c
+++ b/ccan/strgrp/strgrp.c
@@ -103,7 +103,7 @@ strcossim(const int16_t ref[CHAR_N_VALUES], const int16_t key[CHAR_N_VALUES]) {
         sai2 += ref[i] * ref[i];
         sbi2 += key[i] * key[i];
     }
-    return saibi / (sqrt(sai2) * sqrt(sbi2));
+    return 1.0 - (2 * acos(saibi / sqrt(sai2 * sbi2)) / M_PI);
 }
 
 /* Low-cost filter functions */
@@ -117,13 +117,11 @@ should_grp_score_cos(const struct strgrp *const ctx,
 static inline bool
 should_grp_score_len(const struct strgrp *const ctx,
         const struct strgrp_grp *const grp, const char *const str) {
-    const size_t strl = strlen(str);
-    const size_t keyl = grp->key_len;
-    double sr =  strl / keyl;
-    if (1 < sr) {
-        sr = 1 / sr;
-    }
-    return ctx->threshold <= sr;
+    const double lstr = (double) strlen(str);
+    const double lkey = (double) grp->key_len;
+    const double lmin = (lstr > lkey) ? lkey : lstr;
+    const double s = sqrt((2 * lmin * lmin) / (1.0 * lstr * lstr + lkey * lkey));
+    return ctx->threshold <= s;
 }
 
 /* Scoring - Longest Common Subsequence[2]
@@ -147,9 +145,9 @@ lcs(const char *const a, const char *const b) {
     int ia, ib;
     for (ia = (strlen(a) - 1); ia >= 0; ia--) {
         const char iav = a[ia];
+        const int ial = (ia + 1) & 1; // ia last
         for (ib = lb - 1; ib >= 0; ib--) {
             const char ibv = b[ib];
-            const int ial = (ia + 1) & 1; // ia last
             const int iac = ia & 1; // ia current
             const int ibl = ib + 1; // ib last
             // don't need separate "ib current" as it's just ib
@@ -172,7 +170,10 @@ lcs(const char *const a, const char *const b) {
 static inline double
 nlcs(const char *const a, const char *const b) {
     const double lcss = lcs(a, b);
-    return 2 * lcss / (strlen(a) + strlen(b));
+    const double la = (double) strlen(a);
+    const double lb = (double) strlen(b);
+    const double s = sqrt((2 * lcss * lcss) / (la * la + lb * lb));
+    return s;
 }
 
 static inline double