Nlp 在Watson Knowledge Studio中为自定义实体类型创建json文件
我正在尝试为WKS实例上载一组自定义实体类型和子类型 这是本节中的WKS界面视图,您可以在其中定义实体和子实体 upload按钮请求一个json文件 我之前手工创建了一个集合,并下载了json文件 它的第一行如下所示:Nlp 在Watson Knowledge Studio中为自定义实体类型创建json文件,nlp,ibm-watson,watson-knowledge-studio,Nlp,Ibm Watson,Watson Knowledge Studio,我正在尝试为WKS实例上载一组自定义实体类型和子类型 这是本节中的WKS界面视图,您可以在其中定义实体和子实体 upload按钮请求一个json文件 我之前手工创建了一个集合,并下载了json文件 它的第一行如下所示: {"entityTypes":[{**"id":"78361798-b77e-4728-9b6a-f56539c12bcd"**,"label":"Calificativo","sireProp":{"mentionType":null,"subtypes":["Bueno_e
{"entityTypes":[{**"id":"78361798-b77e-4728-9b6a-f56539c12bcd"**,"label":"Calificativo","sireProp":{"mentionType":null,"subtypes":["Bueno_extremo","Bueno_moderado","Regular","Malo_moderado","Malo_extremo"],"roles":["78361798-b77e-4728-9b6a-f56539c12bcd"],"clazz":null,"color":null,"hotkey":null,"backGroundColor":null,"active":true,"roleOnly":false},"creationDate":1583241330349,"source":null,"modifiedDate":0,"typeType":null,"typeClass":null,"typeVersion":null,"typeDesc":null,"typeSuperType":null,"typeSuperTypeId":null,"typeCreateDate":null,"typeUpdateDate":null,"typeProvenance":null,"alchemyAPITypes":null,"nluAPITypes":null},{**"id":"daecb92b-0ce7-4a47-942a-68b50d0cb2fd"**,"label":"TV","sireProp":{"mentionType":null,"subtypes":["Decodificador","Servicio_de_tv"],"roles":
一般来说,内容结构是清晰的,但实体集和内容集都有ID
我想知道是否有办法提前知道,或者生成这些ID,这样我就可以生成包含我想要使用的类型和子类型的整个json,然后上传它
我尝试使用“”代替ID,但收到错误消息,不允许上载。WKS不支持导入自定义json文件,这些文件与根据从WKS工作区导出的文件不同。然而,据我所知,UUID可以是一个有效的
id
字段值,由下面的bash命令生成
$ uuidgen | tr '[:upper:]' '[:lower:]'
此Python脚本生成一个json文件,其格式为WKS所理解:
import uuid
import json
# Generate IDs
ent_id,lbl01_id = uuid.uuid4(), uuid.uuid4()
json_out = {}
json_out.update({
"entityTypes":[{
"id":str(lbl01_id), "label":"Calificativo",
"sireProp":
{
"mentionType":None,
"subtypes":["Bueno_extremo", "Bueno_moderado", "Regular", "Malo_moderado", "Malo_extremo"],
"roles":[str(lbl01_id)], "clazz":None, # Roles relates to self & other labels, if any
"color":None, "hotkey":None, "backGroundColor":None, "active":True, "roleOnly":False
},
"creationDate":1583241330349, "source":None, "modifiedDate":1583247016579, "typeType":None,
"typeClass":None, "typeVersion":None, "typeDesc":None, "typeSuperType":None, "typeSuperTypeId":None,
"typeCreateDate":None, "typeUpdateDate":None, "typeProvenance":None, "alchemyAPITypes":None,
"nluAPITypes":None
}],
"sireInfo":{
"entityProp":{
"mentionType":[{"color":"white", "hotkey":"1", "backGroundColor":"#AA00FF", "name":"NAM"},
{"color":"black", "hotkey":"2", "backGroundColor":"#00FF7F", "name":"NOM"},
{"color":"black", "hotkey":"3", "backGroundColor":"#AAFFFF", "name":"PRO"},
{"color":"white", "hotkey":"4", "backGroundColor":"gray", "name":"NONE"}],
"subtypes":None,
"roles":None,
"clazz":[{"color":"#A5A5A5", "hotkey":"3", "backGroundColor":"white", "name":"SPC"},
{"color":"black", "hotkey":"2", "backGroundColor":"#00FF7F", "name":"NEG"},
{"color":"black", "hotkey":"1", "backGroundColor":"#AAFFFF", "name":"GEN"}],
"color":None,
"hotkey":None,
"backGroundColor":None,
"active":True,
"roleOnly":False
},
"relationProp":{
"tense":[{"name":"PAST"}, {"name":"PRESENT"}, {"name":"FUTURE"}, {"name":"UNSPECIFIED"}],
"modality":[{"name":"ASSERTED"}, {"name":"OTHER"}],
"clazz":[{"name":"SPECIFIC"}, {"name":"NEG"}, {"name":"OTHER"}],
"backGroundColor":None, "color":None, "hotkey":None, "active":True}
},
"functionalEntityTypes":[
{"id":"CATCH_ALL_ENTITY_ID", "label":"*",
"sireProp":{
"mentionType":None, "subtypes":None, "roles":None, "clazz":None, "color":None,
"hotkey":None, "backGroundColor":None, "active":True, "roleOnly":False},
"creationDate":1487227572757, "source":None, "modifiedDate":0, "typeType":None,
"typeClass":None, "typeVersion":None, "typeDesc":None, "typeSuperType":None,
"typeSuperTypeId":None, "typeCreateDate":None, "typeUpdateDate":None, "typeProvenance":None,
"alchemyAPITypes":None, "nluAPITypes":None
}],
"pid":str(ent_id), "modified_date":1583247016579, "kgimported":False
})
with open('json_file.json', 'w') as outfile:
json.dump(json_out, outfile)
这只生成1个实体;要生成更多,只需在“id”和“nluAPITypes”之间,添加实体的次数相同
这里也可以包括“关系类型”非常感谢@iryo!!!我使用了uuid库中Python的uuid.uuid4()命令。将在一分钟内共享完整的代码。