From 5a2be87f658d01bc5b3b7ecdba92ac69bf09dcf7 Mon Sep 17 00:00:00 2001
From: ai_xiaopei <xiaopei@aisim.cn>
Date: Thu, 03 Sep 2026 15:15:37 +0800
Subject: [PATCH] fix(search): 搜索准确性四连修——FTS5 MATCH 引号包裹(连字符不再被解析成 MINUS)；KeywordSearch 截断按命中关键词数排序(高频 bigram 不再挤出强相关文档)；原词奖励档(完整短语压过泛化 bigram)；tags 字段真实参与评分(此前误用 Section)
---
bin/kb-cli | 0
internal/index/fts.go | 59 ++++++++++++++++++++++-------
internal/search/engine.go | 38 +++++++++++++++++-
internal/search/scorer.go | 7 +++
4 files changed, 86 insertions(+), 18 deletions(-)
diff --git a/bin/kb-cli b/bin/kb-cli
index 9202959..dcef256 100755
--- a/bin/kb-cli
+++ b/bin/kb-cli
Binary files differ
diff --git a/internal/index/fts.go b/internal/index/fts.go
index 552b428..83e7838 100644
--- a/internal/index/fts.go
+++ b/internal/index/fts.go
@@ -4,6 +4,7 @@
"database/sql"
"encoding/json"
"sort"
+ "strings"
)
// FTSResult 全文搜索结果
type FTSResult struct {
@@ -13,6 +14,7 @@
Section string
Rank float64
Aliases []string // 别名（JSON 解析）
+ Tags []string // 标签（JSON 解析）
Status string // 状态（草稿/待确认/跟进中 等参与降权）
}
@@ -134,7 +136,17 @@
return nil, nil
}
seen := make(map[int64]*FTSResult)
+ hitCount := make(map[int64]int) // 每个节点被多少个关键词命中（截断排序用）
+ firstIdx := make(map[int64]int) // 首命中位置（截断平手时的次序依据）
var order []int64
+ addCandidate := func(r *FTSResult) {
+ if _, ok := seen[r.ID]; !ok {
+ seen[r.ID] = r
+ firstIdx[r.ID] = len(order)
+ order = append(order, r.ID)
+ }
+ hitCount[r.ID]++
+ }
// 通道 1: FTS（ASCII 词）
var ascii []string
@@ -144,26 +156,30 @@
}
}
for _, kw := range ascii {
+ // FTS5 查询语法里 "-" 是 MINUS 运算符、"(" 等是语法符，
+ // 不引号包裹会报 "no such column" 被静默跳过（如 tag "fill-curve" 搜不到）。
+ // 用双引号包裹成字符串字面量，内部双引号转义。
+ quoted := `"` + strings.ReplaceAll(kw, `"`, `""`) + `"`
rows, err := s.db.Query(`
- SELECT n.id, n.path, n.title, n.section, fts.rank, n.aliases, n.status
+ SELECT n.id, n.path, n.title, n.section, fts.rank, n.aliases, n.tags, n.status
FROM nodes_fts fts JOIN nodes n ON n.id = fts.rowid
- WHERE nodes_fts MATCH ? ORDER BY fts.rank LIMIT ?`, kw, limit)
+ WHERE nodes_fts MATCH ? ORDER BY fts.rank LIMIT ?`, quoted, limit)
if err != nil {
continue // 语法异常词跳过
}
for rows.Next() {
var r FTSResult
- var aliasesJSON string
- if err := rows.Scan(&r.ID, &r.Path, &r.Title, &r.Section, &r.Rank, &aliasesJSON, &r.Status); err != nil {
+ var aliasesJSON, tagsJSON string
+ if err := rows.Scan(&r.ID, &r.Path, &r.Title, &r.Section, &r.Rank, &aliasesJSON, &tagsJSON, &r.Status); err != nil {
continue
}
if aliasesJSON != "" {
_ = json.Unmarshal([]byte(aliasesJSON), &r.Aliases)
}
- if _, ok := seen[r.ID]; !ok {
- seen[r.ID] = &r
- order = append(order, r.ID)
+ if tagsJSON != "" {
+ _ = json.Unmarshal([]byte(tagsJSON), &r.Tags)
}
+ addCandidate(&r)
}
rows.Close()
}
@@ -175,7 +191,7 @@
}
pat := "%" + kw + "%"
rows, err := s.db.Query(`
- SELECT id, path, title, section, aliases, status FROM nodes
+ SELECT id, path, title, section, aliases, tags, status FROM nodes
WHERE title LIKE ? OR aliases LIKE ? OR content_fts LIKE ? OR tags LIKE ?`,
pat, pat, pat, pat)
if err != nil {
@@ -183,24 +199,37 @@
}
for rows.Next() {
var r FTSResult
- var aliasesJSON string
- if err := rows.Scan(&r.ID, &r.Path, &r.Title, &r.Section, &aliasesJSON, &r.Status); err != nil {
+ var aliasesJSON, tagsJSON string
+ if err := rows.Scan(&r.ID, &r.Path, &r.Title, &r.Section, &aliasesJSON, &tagsJSON, &r.Status); err != nil {
continue
}
if aliasesJSON != "" {
_ = json.Unmarshal([]byte(aliasesJSON), &r.Aliases)
}
- r.Rank = 0 // LIKE 无 rank
- if _, ok := seen[r.ID]; !ok {
- seen[r.ID] = &r
- order = append(order, r.ID)
+ if tagsJSON != "" {
+ _ = json.Unmarshal([]byte(tagsJSON), &r.Tags)
}
+ r.Rank = 0 // LIKE 无 rank
+ addCandidate(&r)
}
rows.Close()
}
+ // 截断排序：命中关键词多的节点优先保留，平手按首命中先后。
+ // 旧实现按首命中顺序直接截断，高频泛词（如 bigram "电子" 命中 486 行）
+ // 会把排在后面但与多词强相关的文档挤出候选集。
+ ranked := make([]int64, len(order))
+ copy(ranked, order)
+ sort.SliceStable(ranked, func(i, j int) bool {
+ a, b := ranked[i], ranked[j]
+ if hitCount[a] != hitCount[b] {
+ return hitCount[a] > hitCount[b]
+ }
+ return firstIdx[a] < firstIdx[b]
+ })
+
results := make([]FTSResult, 0, len(order))
- for _, id := range order {
+ for _, id := range ranked {
results = append(results, *seen[id])
}
if limit > 0 && len(results) > limit {
diff --git a/internal/search/engine.go b/internal/search/engine.go
index 13e2ff1..b21224d 100644
--- a/internal/search/engine.go
+++ b/internal/search/engine.go
@@ -81,6 +81,33 @@
textScore[r.ID] = score
}
+ // 原词奖励：候选由 bigram 召回时可能并不含用户原词（"无法"×10 压过
+ // 含"无法开机"整词的文档）。原词真实出现在正文/标题/标签中时给强信号加分，
+ // 使"标题=原词"或"tags 含原词"的文档稳定压过只含泛化 bigram 的文档。
+ for _, r := range candidates {
+ content, _, _, err := store.GetNodeContent(r.ID)
+ if err != nil {
+ continue
+ }
+ lc := strings.ToLower(content)
+ titleL := strings.ToLower(r.Title)
+ for _, kw := range keywords {
+ kwL := strings.ToLower(kw)
+ hit := strings.Contains(lc, kwL) || strings.Contains(titleL, kwL)
+ if !hit {
+ for _, t := range r.Tags {
+ if strings.Contains(strings.ToLower(t), kwL) {
+ hit = true
+ break
+ }
+ }
+ }
+ if hit {
+ textScore[r.ID] += CalcScore(kw, "content", ScoreOriginal)
+ }
+ }
+ }
+
// 3. RWR 图质量（种子 = 候选前 20，全量加载邻接后按种子收敛）
seedIDs := make([]int64, 0, len(candidates))
for _, r := range candidates {
@@ -203,9 +230,14 @@
score += CalcScore(keyword, "title", scoreType)
}
- // 板块匹配
- if strings.Contains(strings.ToLower(r.Section), kw) {
- score += CalcScore(keyword, "tag", scoreType)
+ // 标签匹配（tags 字段真实参与评分；此前误用 Section，而 Section 恒为"其他"）
+ if len(r.Tags) > 0 {
+ for _, t := range r.Tags {
+ if strings.Contains(strings.ToLower(t), kw) {
+ score += CalcScore(keyword, "tag", scoreType)
+ break
+ }
+ }
}
// 别名匹配（命中按 title 档计权，一次命中即计）
diff --git a/internal/search/scorer.go b/internal/search/scorer.go
index 03c2e17..a3b4e3f 100644
--- a/internal/search/scorer.go
+++ b/internal/search/scorer.go
@@ -47,6 +47,7 @@
ScoreNormal ScoreType = iota // 普通词
ScoreExpanded // 扩展词
ScoreSymptom // 症状词
+ ScoreOriginal // 用户原词（完整短语，强信号，高于 bigram 扩展档）
)
// ScoreTable 评分表
@@ -57,6 +58,12 @@
"tag": 2,
"content": 1,
},
+ ScoreOriginal: {
+ "path": 60,
+ "title": 60,
+ "tag": 60,
+ "content": 60,
+ },
ScoreExpanded: {
"path": 40,
"title": 30,
--
Gitblit v1.10.0