|
@@ -91,41 +91,41 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
qu.Catch()
|
|
|
for field, tmps := range result {
|
|
|
for tmpsindex, tmpsvalue := range tmps {
|
|
|
+ if tmpsvalue.ExtFrom == "title" { //标题打分初始化
|
|
|
+ tmps[tmpsindex].Score += TitleScore
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "title初始化", Code: "fieldscore.title", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: TitleScore})
|
|
|
+ }
|
|
|
//是否有段标签
|
|
|
if len(tmpsvalue.BlockTag) > 0 {
|
|
|
//有标签段
|
|
|
var qz float64 = 0.0 //取权重最高的
|
|
|
- var tgk string
|
|
|
for key := range tmpsvalue.BlockTag {
|
|
|
//key = "其他"//TODO 测试用
|
|
|
if TagConfig[key][field] > qz {
|
|
|
qz = TagConfig[key][field]
|
|
|
- tgk = key
|
|
|
}
|
|
|
}
|
|
|
tmps[tmpsindex].Score += 2 * qz //乘以权重系数
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "匹配段标签权重", Code: "权重系数乘以2", RuleText: "BlockTag", Type: tgk + field, ExtFrom: "tagscore.json", Value: tmpsvalue.Value, Score: 2 * qz})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "匹配段标签权重", Code: "权重系数乘以2", RuleText: "BlockTag", ScoreFrom: "tagscore.json", Value: tmpsvalue.Value, Score: 2 * qz})
|
|
|
} else {
|
|
|
//没有段标签,走其他
|
|
|
//qz := TagConfig["其他"][field]
|
|
|
//tmps[tmpsindex].Score += 2 * qz //乘以权重系数
|
|
|
}
|
|
|
- //是否有kv值
|
|
|
+
|
|
|
+ //抽取类型打分
|
|
|
if strings.Contains(tmpsvalue.Type, "colon") {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(SoreConfig["extractype"]["colon"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "colonkv", Code: "fieldscore.colon", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), Type: "colonkv", ExtFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["colon"])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "colonkv", Code: "fieldscore.colon", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["colon"])})
|
|
|
} else if strings.Contains(tmpsvalue.Type, "space") {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(SoreConfig["extractype"]["space"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "spacekv", Code: "fieldscore.space", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), Type: "spacekv", ExtFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["space"])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "spacekv", Code: "fieldscore.space", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["space"])})
|
|
|
} else if strings.Contains(tmpsvalue.Type, "table") {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(SoreConfig["extractype"]["table"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "tablekv", Code: "fieldscore.table", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), Type: "tablekv", ExtFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["table"])})
|
|
|
- }
|
|
|
- if tmpsvalue.ExtFrom != "title" { //非标题抽取
|
|
|
- if strings.Contains(tmpsvalue.Type, "regexp") {
|
|
|
- tmps[tmpsindex].Score += qu.Float64All(SoreConfig["extractype"]["regexp"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "regexp", Code: "fieldscore.regexp", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), Type: "regexp", ExtFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["regexp"])})
|
|
|
- }
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "tablekv", Code: "fieldscore.table", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["table"])})
|
|
|
+ } else if strings.Contains(tmpsvalue.Type, "regexp") {
|
|
|
+ tmps[tmpsindex].Score += qu.Float64All(SoreConfig["extractype"]["regexp"])
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "regexp", Code: "fieldscore.regexp", RuleText: qu.ObjToString(SoreConfig["extractype"]["describe"]), ScoreFrom: "fieldscore.json", Value: tmpsvalue.Value, Score: qu.Float64All(SoreConfig["extractype"]["regexp"])})
|
|
|
}
|
|
|
scoreRule := SoreConfig[field]
|
|
|
if scoreRule == nil {
|
|
@@ -139,7 +139,7 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
}
|
|
|
if valueLen > 100 && field != "projectscope" {
|
|
|
tmps[tmpsindex].Score = -99
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: `valueLen > 100 && field != "projectscope"直接-99分`, Code: field, Type: "length", Value: tmpsvalue.Value, Score: -99})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: `valueLen > 100 && field != "projectscope"直接-99分`, Code: field, Value: tmpsvalue.Value, Score: -99})
|
|
|
}
|
|
|
if lengths, ok := scoreRule["length"].([]interface{}); ok {
|
|
|
for _, tmp := range lengths {
|
|
@@ -152,13 +152,13 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
}
|
|
|
if valueLen < min {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[0])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "长度打分", Code: fmt.Sprint(valueLen, "<", min), Type: field, ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[0])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "长度打分", Code: fmt.Sprint(valueLen, "<", min), ScoreFrom: "fieldscore.json.length", Value: tmpsvalue.Value, Score: qu.Float64All(scores[0])})
|
|
|
} else if valueLen > max {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[2])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "长度打分", Code: fmt.Sprint(valueLen, ">", max), Type: field, ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[2])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "长度打分", Code: fmt.Sprint(valueLen, ">", max), ScoreFrom: "fieldscore.json.length", Value: tmpsvalue.Value, Score: qu.Float64All(scores[2])})
|
|
|
} else {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[1])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "长度打分", Code: fmt.Sprint(valueLen, ">", min, "&&", valueLen, "<", max), Type: field, ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[1])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "长度打分", Code: fmt.Sprint(valueLen, ">", min, "&&", valueLen, "<", max), ScoreFrom: "fieldscore.json.length", Value: tmpsvalue.Value, Score: qu.Float64All(scores[1])})
|
|
|
}
|
|
|
}
|
|
|
}
|
|
@@ -172,7 +172,7 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
reg := p["regexp"].(*regexp.Regexp)
|
|
|
if reg.MatchString(qu.ObjToString(tmpsvalue.Value)) {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(p["score"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "负面词打分" + fmt.Sprint(p["describe"]), Code: field + ".negativewords", RuleText: reg.String(), Type: "regexp", ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(p["score"])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "负面词打分" + fmt.Sprint(p["describe"]), Code: "negativewords", RuleText: reg.String(), ScoreFrom: "fieldscore.json.negativewords", Value: tmpsvalue.Value, Score: qu.Float64All(p["score"])})
|
|
|
}
|
|
|
}
|
|
|
}, func(err interface{}) {
|
|
@@ -190,7 +190,7 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
reg := p["regexp"].(*regexp.Regexp)
|
|
|
if reg.MatchString(qu.ObjToString(tmpsvalue.Value)) {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(p["score"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "正面词打分" + fmt.Sprint(p["describe"]), Code: field + ".positivewords", RuleText: reg.String(), Type: "regexp", ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(p["score"])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "正面词打分" + fmt.Sprint(p["describe"]), Code: "positivewords", RuleText: reg.String(), ScoreFrom: "fieldscore.json.positivewords", Value: tmpsvalue.Value, Score: qu.Float64All(p["score"])})
|
|
|
}
|
|
|
}
|
|
|
}, func(err interface{}) {
|
|
@@ -208,7 +208,7 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
reg := p["regexp"].(*regexp.Regexp)
|
|
|
if reg.MatchString(qu.ObjToString(tmpsvalue.Value)) {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(p["score"])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "位置打分winnerorder" + fmt.Sprint(p["describe"]), Code: field + ".winnerorder", RuleText: reg.String(), Type: "regexp", ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(p["score"])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "位置打分winnerorder" + fmt.Sprint(p["describe"]), Code: "winnerorder", RuleText: reg.String(), ScoreFrom: "fieldscore.json.winnerorder", Value: tmpsvalue.Value, Score: qu.Float64All(p["score"])})
|
|
|
}
|
|
|
}
|
|
|
}, func(err interface{}) {
|
|
@@ -229,13 +229,13 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
}
|
|
|
if val < min && 0 < val {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[0])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "数据范围打分", Code: field + ".float", RuleText: fmt.Sprint(val, "<", min, "&&", 0, "<", val), ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[0])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "数据范围打分", Code: field + ".float", RuleText: fmt.Sprint(val, "<", min, "&&", 0, "<", val), ScoreFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[0])})
|
|
|
} else if val > max {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[2])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "数据范围打分", Code: field + ".float", RuleText: fmt.Sprint(val, ">", max), ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[2])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "数据范围打分", Code: field + ".float", RuleText: fmt.Sprint(val, ">", max), ScoreFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[2])})
|
|
|
} else if val <= max && val >= min {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[1])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "数据范围打分", Code: field + ".float", RuleText: fmt.Sprintln(val, "<=", max, "&&", val, ">=", min), ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[1])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "数据范围打分", Code: field + ".float", RuleText: fmt.Sprintln(val, "<=", max, "&&", val, ">=", min), ScoreFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[1])})
|
|
|
}
|
|
|
}
|
|
|
//其他打分配置
|
|
@@ -250,10 +250,10 @@ func ScoreFields(j *ju.Job) map[string][]*ju.ExtField {
|
|
|
}
|
|
|
if val > max {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[2])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "其他打分配置decimal", Code: field + ".decimal", RuleText: fmt.Sprint(val, ">", max), ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[2])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "其他打分配置decimal", Code: field + ".decimal", RuleText: fmt.Sprint(val, ">", max), ScoreFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[2])})
|
|
|
} else if val <= max && val > min {
|
|
|
tmps[tmpsindex].Score += qu.Float64All(scores[1])
|
|
|
- tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "其他打分配置decimal", Code: field + ".decimal", RuleText: fmt.Sprint(val, "<=", max, "&&", val, ">", min), ExtFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[1])})
|
|
|
+ tmps[tmpsindex].ScoreItem = append(tmps[tmpsindex].ScoreItem, &ju.ScoreItem{Des: "其他打分配置decimal", Code: field + ".decimal", RuleText: fmt.Sprint(val, "<=", max, "&&", val, ">", min), ScoreFrom: "fieldscore.json." + field, Value: tmpsvalue.Value, Score: qu.Float64All(scores[1])})
|
|
|
}
|
|
|
}
|
|
|
}
|