summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorPeter Wu <peter@lekensteyn.nl>2015-04-02 11:41:02 +0200
committerPeter Wu <peter@lekensteyn.nl>2015-04-02 11:41:02 +0200
commita9839f2a780c9ddf117ad3cf39daadd5333112ee (patch)
tree51e914197d31614f667645534abb939a87631e56
parentbe34f8b98aa4942953deb383ea61156ee1bf20b1 (diff)
downloadassignment4-a9839f2a780c9ddf117ad3cf39daadd5333112ee.tar.gz
bigram smooth tweak
-rw-r--r--spellchecker/src/CorpusReader.java3
1 files changed, 2 insertions, 1 deletions
diff --git a/spellchecker/src/CorpusReader.java b/spellchecker/src/CorpusReader.java
index 8281210..686f243 100644
--- a/spellchecker/src/CorpusReader.java
+++ b/spellchecker/src/CorpusReader.java
@@ -119,7 +119,8 @@ public class CorpusReader {
// probability as result.
if (NGram.indexOf(' ') != -1) {
// bigram, must be the nominator
- smoothedCount += 1;
+ // we do not have a lot of bigrams
+ smoothedCount += .01;
} else {
// unigram, must be the denominator
smoothedCount += 1;