From b2e05a1d41540586cbaa7c8ee532393501601154 Mon Sep 17 00:00:00 2001
From: alex <alexander.vrabel@proton.me>
Date: Wed, 29 Jul 2026 00:56:57 +0200
Subject: added proximity based scoring

---
 .gitignore                |  2 +-
 core/index/index.go       | 51 +++++++++++++++++++++++++++++++++++++----------
 core/scoring/proximity.go | 31 ++++++++++++++++++++++++++++
 core/search/search.go     | 15 ++++++++++++++
 4 files changed, 88 insertions(+), 11 deletions(-)
 create mode 100644 core/scoring/proximity.go

diff --git a/.gitignore b/.gitignore
index 5f28270..4e71595 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1 @@
-
\ No newline at end of file
+wikipedia_index.gob
\ No newline at end of file
diff --git a/core/index/index.go b/core/index/index.go
index 2e669fc..b00f524 100644
--- a/core/index/index.go
+++ b/core/index/index.go
@@ -1,6 +1,7 @@
 package index
 
 import (
+	"bufio"
 	"encoding/gob"
 	"os"
 	"searchEngine/core/analyzer"
@@ -14,9 +15,10 @@ type Posting struct {
 }
 
 type InvertedIndex struct {
-	Mu       sync.RWMutex `gob:"-"`
-	Data     map[string][]Posting
-	DocNames map[int]string
+	Mu         sync.RWMutex `gob:"-"`
+	Data       map[string][]Posting
+	DocNames   map[int]string
+	DocLengths map[int]int
 }
 
 func New() *InvertedIndex {
@@ -51,7 +53,7 @@ func (idx *InvertedIndex) Add(docID int, title string, bodyTokens []string) {
 	defer idx.Mu.Unlock()
 
 	idx.DocNames[docID] = title
-
+	idx.DocLengths[docID] = len(bodyTokens)
 	for token, stats := range docStats {
 		idx.Data[token] = append(idx.Data[token], Posting{
 			DocId:          docID,
@@ -71,8 +73,25 @@ func (idx *InvertedIndex) Save(filepath string) error {
 	}
 	defer file.Close()
 
-	encoder := gob.NewEncoder(file)
-	return encoder.Encode(idx)
+	writer := bufio.NewWriter(file)
+
+	dto := struct {
+		Data       map[string][]Posting
+		DocNames   map[int]string
+		DocLengths map[int]int
+	}{
+		Data:       idx.Data,
+		DocNames:   idx.DocNames,
+		DocLengths: idx.DocLengths,
+	}
+
+	encoder := gob.NewEncoder(writer)
+	err = encoder.Encode(dto)
+
+	if flushErr := writer.Flush(); flushErr != nil {
+		return flushErr
+	}
+	return err
 }
 
 func Load(filepath string) (*InvertedIndex, error) {
@@ -82,11 +101,23 @@ func Load(filepath string) (*InvertedIndex, error) {
 	}
 	defer file.Close()
 
-	idx := &InvertedIndex{}
-	decoder := gob.NewDecoder(file)
-	err = decoder.Decode(idx)
+	reader := bufio.NewReader(file)
+
+	var dto struct {
+		Data       map[string][]Posting
+		DocNames   map[int]string
+		DocLengths map[int]int
+	}
+
+	decoder := gob.NewDecoder(reader)
+	err = decoder.Decode(&dto)
 	if err != nil {
 		return nil, err
 	}
-	return idx, nil
+
+	return &InvertedIndex{
+		Data:       dto.Data,
+		DocNames:   dto.DocNames,
+		DocLengths: dto.DocLengths,
+	}, nil
 }
diff --git a/core/scoring/proximity.go b/core/scoring/proximity.go
new file mode 100644
index 0000000..f765aa7
--- /dev/null
+++ b/core/scoring/proximity.go
@@ -0,0 +1,31 @@
+package scoring
+
+import "math"
+
+func CalculateProximity(pos1, pos2 []int) float64 {
+	var score float64
+	i, j := 0, 0
+
+	for i < len(pos1) && j < len(pos2) {
+		p1 := pos1[i]
+		p2 := pos2[j]
+
+		dist := math.Abs(float64(p1 - p2))
+
+		if dist == 1 {
+			score += 15.0
+		} else if dist <= 5 {
+			score += 5.0
+		} else if dist <= 15 {
+			score += 1.0
+		}
+
+		if p1 < p2 {
+			i++
+		} else {
+			j++
+		}
+	}
+
+	return score
+}
diff --git a/core/search/search.go b/core/search/search.go
index 721574c..cf1054b 100644
--- a/core/search/search.go
+++ b/core/search/search.go
@@ -24,6 +24,7 @@ func Search(idx *index.InvertedIndex, query string) []string {
 
 	totalDocs := len(idx.DocNames)
 	docMatchCounts := make(map[int]int)
+	docPostings := make(map[int][]index.Posting)
 	scores := make(map[int]float64)
 
 	uniqueMap := make(map[string]bool)
@@ -48,6 +49,8 @@ func Search(idx *index.InvertedIndex, query string) []string {
 			docMatchCounts[p.DocId]++
 			termScore := scoring.ScoreTFIDF(p, idf)
 			scores[p.DocId] += termScore
+
+			docPostings[p.DocId] = append(docPostings[p.DocId], p)
 		}
 	}
 
@@ -57,7 +60,19 @@ func Search(idx *index.InvertedIndex, query string) []string {
 			results = append(results, Result{DocID: docID, Score: scores[docID]})
 		}
 	}
+	for i := range results {
+		proximityBonus := 0.0
+		resultsPostings := docPostings[results[i].DocID]
+		for j := 0; j < len(resultsPostings)-1; j++ {
+			bodyBonus := scoring.CalculateProximity(resultsPostings[j].BodyPositions, resultsPostings[j+1].BodyPositions)
+
+			titleBonus := scoring.CalculateProximity(resultsPostings[j].TitlePositions, resultsPostings[j+1].TitlePositions)
 
+			proximityBonus += bodyBonus + (titleBonus * 5.0)
+		}
+
+		results[i].Score += proximityBonus
+	}
 	slices.SortFunc(results, func(a, b Result) int {
 		return cmp.Compare(b.Score, a.Score)
 	})
-- 
cgit v1.2.3