wangweiye
/
mathdeptv2
Archived
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

增添几个题目相同与无关标识

This commit is contained in:
weiye.wang 2023-04-26 23:08:03 +08:00
parent 2fbc5ca20f
commit 2cd90643dc
2 changed files with 94 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
工具/相同相似题目标注.py
View File

@ -1,6 +1,7 @@
import os,re,json import os,re,json
filename = "临时文件/相似题目.txt" # filename = "临时文件/相似题目.txt"
filename = "临时文件/相似1.txt"
# 读取题库数据并转换为字典 # 读取题库数据并转换为字典
with open(r"../题库0.3/Problems.json","r",encoding = "utf8") as f: with open(r"../题库0.3/Problems.json","r",encoding = "utf8") as f:
@ -12,7 +13,11 @@ with open(filename,"r",encoding = "utf8") as f:
similar_text = "\n"+f.read() similar_text = "\n"+f.read()
similar_types = re.findall(r"\n[\d]\.[\d]{4}[\s]*([srSRnN])*[\s]*\n",similar_text) similar_types = re.findall(r"\n[\d]\.[\d]{4}[\s]*([srSRnN])*[\s]*\n",similar_text)
similar_problems = re.findall(r"\n([\d]{6}) ",similar_text) similar_problems = re.findall(r"\n([\d]{6})\s",similar_text)
print(similar_types)
print(similar_problems)
samecount ,relcount, unrelcount = 0,0,0
if len(similar_types) * 2 == len(similar_problems): if len(similar_types) * 2 == len(similar_problems):
for i in similar_types: for i in similar_types:
@ -21,19 +26,40 @@ if len(similar_types) * 2 == len(similar_problems):
if i.upper() == "S": if i.upper() == "S":
if not id2 in pro_dict[id1]["same"]: if not id2 in pro_dict[id1]["same"]:
pro_dict[id1]["same"].append(id2) pro_dict[id1]["same"].append(id2)
samecount += 1
print("相同题目已标注: %s <- %s"%(id1,id2))
if not id1 in pro_dict[id2]["same"]: if not id1 in pro_dict[id2]["same"]:
pro_dict[id2]["same"].append(id1) pro_dict[id2]["same"].append(id1)
print("相同题目已标注:",id1,id2) samecount += 1
print("相同题目已标注: %s <- %s"%(id2,id1))
elif i.upper() == "R": elif i.upper() == "R":
if not id2 in pro_dict[id1]["related"]: if not id2 in pro_dict[id1]["related"]:
pro_dict[id1]["related"].append(id2) pro_dict[id1]["related"].append(id2)
relcount += 1
print("关联题目已标注: %s <- %s"%(id1,id2))
if not id1 in pro_dict[id2]["related"]: if not id1 in pro_dict[id2]["related"]:
pro_dict[id2]["related"].append(id1) pro_dict[id2]["related"].append(id1)
print("关联题目已标注:",id1,id2) relcount += 1
print("关联题目已标注: %s <- %s"%(id2,id1))
elif i.upper() == "N":
if not id2 in pro_dict[id1]["unrelated"]:
pro_dict[id1]["unrelated"].append(id2)
unrelcount += 1
print("无关题目已标注: %s <- %s"%(id1,id2))
if not id1 in pro_dict[id2]["unrelated"]:
pro_dict[id2]["unrelated"].append(id1)
unrelcount += 1
print("无关题目已标注: %s <- %s"%(id2,id1))
else: else:
print("相似程度数据:",len(similar_types),"个, 相似题目:",len(similar_problems),"题. 数据有问题, 请检查.") print("相似程度数据:",len(similar_types),"个, 相似题目:",len(similar_problems),"题. 数据有问题, 请检查.")
print("已标注相同题目 %d 项, 关联题目 %d 项, 无关题目 %d 项"%(samecount,relcount,unrelcount))
# 将题库字典转换为json文件并保存至原位 # 将题库字典转换为json文件并保存至原位
database = json.dumps(pro_dict,indent=4,ensure_ascii=False) database = json.dumps(pro_dict,indent=4,ensure_ascii=False)

85
题库0.3/Problems.json
View File

@ -132818,7 +132818,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012013"
],
"related": [ "related": [
"004447", "004447",
"005789", "005789",
@ -132846,7 +132848,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012014"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -132873,7 +132877,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012015"
],
"related": [ "related": [
"000808" "000808"
], ],
@ -132899,7 +132905,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012016"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -132925,7 +132933,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012017"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -132950,7 +132960,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012018"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -132976,7 +132988,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012019"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -133042,7 +133056,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012021"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -133066,7 +133082,9 @@
"edit": [ "edit": [
"20220711\t王伟叶" "20220711\t王伟叶"
], ],
"same": [], "same": [
"012022"
],
"related": [ "related": [
"003601" "003601"
], ],
@ -265665,7 +265683,10 @@
"related": [], "related": [],
"remark": "", "remark": "",
"space": "4em", "space": "4em",
"unrelated": [] "unrelated": [
"011798",
"011799"
]
}, },
"010055": { "010055": {
"id": "010055", "id": "010055",
@ -309684,7 +309705,9 @@
], ],
"remark": "", "remark": "",
"space": "4em", "space": "4em",
"unrelated": [] "unrelated": [
"010054"
]
}, },
"011799": { "011799": {
"id": "011799", "id": "011799",
@ -309708,7 +309731,9 @@
], ],
"remark": "", "remark": "",
"space": "4em", "space": "4em",
"unrelated": [] "unrelated": [
"010054"
]
}, },
"011800": { "011800": {
"id": "011800", "id": "011800",
@ -314629,7 +314654,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004665"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314658,7 +314685,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004666"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314687,7 +314716,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004667"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314716,7 +314747,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004668"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314745,7 +314778,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004669"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314775,7 +314810,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004670"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314804,7 +314841,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004671"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314873,7 +314912,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004673"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",
@ -314902,7 +314943,9 @@
"edit": [ "edit": [
"20221027\t王伟叶" "20221027\t王伟叶"
], ],
"same": [], "same": [
"004674"
],
"related": [], "related": [],
"remark": "", "remark": "",
"space": "", "space": "",