add comments and fixes compute B in k mixture model

2 files changed, 5 insertions, 1 deletions

diff --git a/utils/training/gen_k_mixture_model.cpp b/utils/training/gen_k_mixture_model.cpp
index c26ac49..13ff04d 100644
--- a/utils/training/gen_k_mixture_model.cpp
+++ b/utils/training/gen_k_mixture_model.cpp
@@ -218,7 +218,7 @@ static bool train_second_word(KMixtureModelBigram * bigram,
         single_gram = new KMixtureModelSingleGram;
     train_single_gram(hash_of_document, single_gram, token1, delta);
 
-    if ( 0 == delta ){
+    if ( 0 == delta ){ /* Please consider maximum occurs allowed. */
         delete single_gram;
         return false;
     }
diff --git a/utils/training/k_mixture_model.h b/utils/training/k_mixture_model.h
index 710fe0b..9dda1c2 100644
--- a/utils/training/k_mixture_model.h
+++ b/utils/training/k_mixture_model.h
@@ -54,6 +54,10 @@ static inline parameter_t compute_B(corpus_count_t N,
                                     corpus_count_t T,
                                     corpus_count_t n_0,
                                     corpus_count_t n_1){
+    /* Please consider B_2 is no less than 2 in paper. */
+    if ( 0 == T - n_1 && 0 == N - n_0 - n_1 )
+        return 2;
+
     parameter_t B = (T - n_1 ) / (parameter_t) (N - n_0 - n_1);
     return B;
 }