begin work on better searching algorithms, created logic for tf-idf

1 files changed, 17 insertions, 0 deletions

diff --git a/scoring/idf.go b/scoring/idf.go
new file mode 100644
index 0000000..6d96089
--- /dev/null
+++ b/scoring/idf.go
@@ -0,0 +1,17 @@
+package scoring
+
+import (
+	"math"
+	"searchEngine/index"
+)
+
+func CalculateIDF(totalDocs int, docFrequency int) float64 {
+	if docFrequency == 0 {
+		return 0.0
+	}
+	return math.Log10(float64(totalDocs) / float64(docFrequency))
+}
+func ScoreTFIDF(p index.Posting, idf float64) float64 {
+
+	return float64(p.Count) * idf
+}