diff options
Diffstat (limited to 'src/classifiers/winnow.c')
-rw-r--r-- | src/classifiers/winnow.c | 15 |
1 files changed, 12 insertions, 3 deletions
diff --git a/src/classifiers/winnow.c b/src/classifiers/winnow.c index 24ee7821c..2e8b98423 100644 --- a/src/classifiers/winnow.c +++ b/src/classifiers/winnow.c @@ -213,7 +213,10 @@ winnow_classify (struct classifier_ctx *ctx, statfile_pool_t * pool, GTree * inp if (ctx->cfg->opts && (value = g_hash_table_lookup (ctx->cfg->opts, "min_tokens")) != NULL) { minnodes = strtol (value, NULL, 10); - nodes = g_tree_nnodes (input) / FEATURE_WINDOW_SIZE; + nodes = g_tree_nnodes (input); + if (nodes > FEATURE_WINDOW_SIZE) { + nodes = nodes / FEATURE_WINDOW_SIZE + FEATURE_WINDOW_SIZE; + } if (nodes < minnodes) { msg_info ("do not classify message as it has too few tokens: %d, while %d min", nodes, minnodes); return FALSE; @@ -305,7 +308,10 @@ winnow_weights (struct classifier_ctx *ctx, statfile_pool_t * pool, GTree * inpu if (ctx->cfg->opts && (value = g_hash_table_lookup (ctx->cfg->opts, "min_tokens")) != NULL) { minnodes = strtol (value, NULL, 10); - nodes = g_tree_nnodes (input) / FEATURE_WINDOW_SIZE; + nodes = g_tree_nnodes (input); + if (nodes > FEATURE_WINDOW_SIZE) { + nodes = nodes / FEATURE_WINDOW_SIZE + FEATURE_WINDOW_SIZE; + } if (nodes < minnodes) { msg_info ("do not classify message as it has too few tokens: %d, while %d min", nodes, minnodes); return NULL; @@ -379,7 +385,10 @@ winnow_learn (struct classifier_ctx *ctx, statfile_pool_t *pool, const char *sym if (ctx->cfg->opts && (value = g_hash_table_lookup (ctx->cfg->opts, "min_tokens")) != NULL) { minnodes = strtol (value, NULL, 10); - nodes = g_tree_nnodes (input) / FEATURE_WINDOW_SIZE; + nodes = g_tree_nnodes (input); + if (nodes > FEATURE_WINDOW_SIZE) { + nodes = nodes / FEATURE_WINDOW_SIZE + FEATURE_WINDOW_SIZE; + } if (nodes < minnodes) { msg_info ("do not learn message as it has too few tokens: %d, while %d min", nodes, minnodes); if (sum != NULL) { |