Browse Source

正式任务city抽取

unknown 6 years ago
parent
commit
1ab3b307f8
3 changed files with 19 additions and 21 deletions
  1. 4 3
      src/jy/extract/extract.go
  2. 9 14
      src/jy/extract/extractInit.go
  3. 6 4
      src/jy/extract/extractcity.go

+ 4 - 3
src/jy/extract/extract.go

@@ -94,9 +94,10 @@ func StartExtractTaskId(taskId string) bool {
 	ext.InitTag()
 	ext.InitClearFn()
 
-	//	ext.InitProvince()
-	//	ext.InitCityAll()
-	//	ext.InitCitySim()
+	ext.InitProvince()
+	ext.InitCityAll()
+	ext.InitCitySim()
+	InitDFA()
 
 	ext.IsRun = true
 	if isgo {

+ 9 - 14
src/jy/extract/extractInit.go

@@ -481,12 +481,7 @@ func InitDFA() {
 			}
 		}
 	}
-	//	log.Println("ProvinceMap---", ProvinceMap)
-	//	log.Println("ProvinceBrief---", ProvinceBrief)
-	//	log.Println("CityBrief---", CityBrief)
-	//	log.Println("AreaToCity---", AreaToCity)
-	//	log.Println("DistrictCityMap---", DistrictCityMap)
-	//	log.Println("StreetDistrictMap---", StreetDistrictMap)
+
 	//加载简称
 	AreaSimGet = DFA{}
 	for k, v := range CitySimConfig {
@@ -535,14 +530,14 @@ func InitDFA() {
 					}
 					AreaToCity[ss] = cs
 
-					d := &District{}
-					d.Name = ss
-					d.C = c
-					AreaDistrict.AddWord(ss) //加入区或县敏感词
-					ctmp := DistrictCityMap[ss]
-					if ctmp == nil {
-						DistrictCityMap[ss] = c
-					}
+					//					d := &District{}
+					//					d.Name = ss
+					//					d.C = c
+					//					AreaDistrict.AddWord(ss) //加入区或县敏感词
+					//					ctmp := DistrictCityMap[ss]
+					//					if ctmp == nil {
+					//						DistrictCityMap[ss] = c
+					//					}
 				}
 			}
 		}

+ 6 - 4
src/jy/extract/extractcity.go

@@ -98,7 +98,8 @@ func ExtractDistrict(field []string, bres bool, c, p, id string) (bool, string,
 							bres = true
 						} else { //对比抽到的城市省份是否一致
 							if c != ctmp || p != ptmp {
-								log.Println("City And Province, Inconsistent Before And After,Id:", id)
+								log.Println(str, word)
+								log.Println("district: City And Province, Inconsistent Before And After,Id:", id, c, p, ctmp, ptmp, d)
 							}
 						}
 					}
@@ -115,7 +116,8 @@ func ExtractDistrict(field []string, bres bool, c, p, id string) (bool, string,
 							bres = true
 						} else { //对比抽到的城市省份是否一致
 							if c != ctmp || p != ptmp {
-								log.Println("City And Province, Inconsistent Before And After,Id:", id)
+								log.Println(str, word)
+								log.Println("street: City And Province, Inconsistent Before And After,Id:", id, c, p, ctmp, ptmp, d)
 							}
 						}
 					}
@@ -132,10 +134,10 @@ func ExtractProvinceCity(province, city, id string, field []string) (bres bool,
 	bc := true //是否继续抽取
 	if city != "" {
 		if CityBrief[city] == nil { //简称不存在
-			log.Println("city err:", city, id)
+			//log.Println("city err:", city, id)
 		} else { //简称存在
 			if province != CityBrief[city].P.Brief { //省份不配对
-				log.Println("province err:", city, province, id)
+				//log.Println("province err:", city, province, id)
 			} else {
 				bc = false
 				//城市省份都正确