Copilot commented on code in PR #271: URL: https://github.com/apache/incubator-hugegraph-ai/pull/271#discussion_r2138019225
########## semi-automated generated extracted prompt/New_semi-automated generated extracted prompt2.py: ########## @@ -0,0 +1,95 @@ +#-*- coding: utf-8 -*- +#coding:utf-8 +import requests +import json +import re +import pandas as pd +from sklearn.feature_extraction.text import TfidfVectorizer +from sklearn.metrics.pairwise import cosine_similarity + +def API(question): Review Comment: The API function appears identical to the one in another file; consider refactoring it into a shared module to reduce duplication. ########## semi-automated generated extracted prompt/New_semi-automated generated extracted prompt1.py: ########## @@ -0,0 +1,81 @@ +#-*- coding: utf-8 -*- +#coding:utf-8 +import requests +import json +import re +import pandas as pd + +def API(question): Review Comment: The API function is duplicated across files; consider consolidating this functionality into a common module to enhance code maintainability. ########## semi-automated generated extracted prompt/New_semi-automated generated extracted prompt2.py: ########## @@ -0,0 +1,95 @@ +#-*- coding: utf-8 -*- +#coding:utf-8 +import requests +import json +import re +import pandas as pd +from sklearn.feature_extraction.text import TfidfVectorizer +from sklearn.metrics.pairwise import cosine_similarity + +def API(question): + url="https://"#API服务 + headers={ + "Authorization": "",#自己的token + "Accept": "application/json", + "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 Chrome/94.0.4606.71 Safari/537.36", + "Origin": "https://";, + "Referer" : "https://"; + } + response = requests.get(url, headers=headers) + print("状态码：", response.status_code) + print("响应内容：", response.text) + + my_chat_id = response.json()["data"] + print("chat_id:", my_chat_id) + + message_url = f"https://{my_chat_id}"; + payload = { + "message": question, + "re_chat": False, + "image_list": [], + "document_list": [], + "audio_list": [], + "video_list": [], + "form_data": {} + } + headers = { + "Authorization": "",#自己的token + "Content-Type": "application/json", + "Accept": "*/*" + } + response = requests.post(message_url, data=json.dumps(payload), headers=headers, stream=True) + #处理流式返回数据 + full_content_list = [] + for line in response.iter_lines(): + if line: + if line.startswith(b"data:"): + content_json = json.loads(line[5:].decode("utf-8")) + full_content_list.append(content_json.get("content", "")) + full_content = "".join(full_content_list) + return full_content + +def Recall_most_similar_template(query,templates): + texts=[query]+templates#把query和现有模版组合并到列表 + vectorizer=TfidfVectorizer() + tfidf_matrix=vectorizer.fit_transform(texts)#向量化 + similarities=cosine_similarity(tfidf_matrix[0:1],tfidf_matrix[1:])#query的TF-IDF向量和模版组的组合，求余弦相似度 + most_similar_idx=similarities.argmax() + return templates[most_similar_idx] + +templates = ['法律知识图谱','人物知识图谱','人物关系图谱','医疗知识图谱','法律知识图谱','企业知识图谱','科研知识图谱','技术图谱','舆情知识图谱','政策知识图谱','法律图谱','地理知识图谱','商业知识图谱','组织知识图谱','知识问答图谱','能力/技能图谱','技术创新图谱','科学知识图谱','金融知识图谱','故事情节图谱','发展脉络图谱'] + +df = pd.read_excel(r'C:\Data_Test.xlsx') +for index,row in df.iterrows(): + UserInput = row['原文']+row['指令'] + + GenerateTemplate4User = f"""用户目标：输入文本和要求，依靠大模型抽取知识图谱。你需要撰写符合用户要求的提示词，用于引导其他大模型根据用户输入文本抽取知识图谱。 + 请按照分类和结构化的方式来回答这个问题： + 1. 分类用户提问的主要部分或类别，明确用户要构建什么知识图谱。用户提问可能在句首或句尾，请在这一步明确输出用户提问原句，在原句左右各用五个加号连接+++++； + 2. 详细分析用户目标，并按照用户目标分析图谱的实体分类、关系分类、每个实体的属性； + 3. 记录中间结果，列出实体列表，关系列表，每个实体的属性列表； + 4. 由上述结论总结得出用于指导大模型生成符合用户要求的Prompt，输出最终版本的提示词，该提示词需要包含第二步第三步中详细信息以及Json格式的示例，输出结果放置在下面括号内<<<<<>>>>> + + 用户输入如下： + {UserInput} + """ + ##回复每个子问题，提取并标记用户要求，据此生成Prompt + result=API(GenerateTemplate4User) + df.loc[index,'分析过程']=result + + ##获取+标记的用户要求 + pattern1 = r'\+{3,}(.*?)\+{3,}' + matches_query=re.findall(pattern1,result,re.DOTALL) + template=Recall_most_similar_template(matches_query[0],templates) + + ##过去模型生成的Prompt + pattern2 = r'<{3,}(.*?)>{3,}' + matches_template=re.findall(pattern2,result,re.DOTALL) + df.loc[index, '生成指令']=matches_template + + # 测试模型生成的Prompt的建图效果 + BuildGraph4User=f"按照下面要求建立知识图谱：{matches_query[-1]}，然后将输出知识图谱转化为Neo4j风格的可执行语句，用户输入如下{UserInput}。该知识图谱类型有可能与{template}的建立方式接近，也有可能无关，你可以完全忽视结尾给出的参考" + result=API(BuildGraph4User) + df.loc[index,'应用该指令效果']=result + + df.to_excel(r'C:\Data_Test.xlsx', index=False) Review Comment: Consider moving the Excel write operation outside of the for loop to reduce unnecessary I/O operations, especially if processing a large dataset. ```suggestion # Write the modified DataFrame to the Excel file after processing all rows df.to_excel(r'C:\Data_Test.xlsx', index=False) ``` ########## semi-automated generated extracted prompt/New_semi-automated generated extracted prompt1.py: ########## @@ -0,0 +1,81 @@ +#-*- coding: utf-8 -*- +#coding:utf-8 +import requests +import json +import re +import pandas as pd + +def API(question): + url="https://"#API服务 + headers={ + "Authorization": "",#自己的token + "Accept": "application/json", + "User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 Chrome/94.0.4606.71 Safari/537.36", + "Origin": "https://";, + "Referer" : "https://"; + } + response = requests.get(url, headers=headers) + print("状态码：", response.status_code) + print("响应内容：", response.text) + + my_chat_id = response.json()["data"] + print("chat_id:", my_chat_id) + + message_url = f"https://{my_chat_id}"; + payload = { + "message": question, + "re_chat": False, + "image_list": [], + "document_list": [], + "audio_list": [], + "video_list": [], + "form_data": {} + } + headers = { + "Authorization": "",#自己的token + "Content-Type": "application/json", + "Accept": "*/*" + } + response = requests.post(message_url, data=json.dumps(payload), headers=headers, stream=True) + #处理流式返回数据 + full_content_list = [] + for line in response.iter_lines(): + if line: + if line.startswith(b"data:"): + content_json = json.loads(line[5:].decode("utf-8")) + full_content_list.append(content_json.get("content", "")) + full_content = "".join(full_content_list) + return full_content + +#读入test_data表 +df = pd.read_excel(r'C:\Data_Test.xlsx') +for index,row in df.iterrows(): + UserInput = row['原文']+row['指令'] + + GenerateTemplate4User = f"""用户目标：输入文本和要求，依靠大模型抽取知识图谱。你需要撰写符合用户要求的提示词，用于引导其他大模型根据用户输入文本抽取知识图谱。 + 请按照分类和结构化的方式来回答这个问题： + 1. 分类用户提问的主要部分或类别，明确用户要构建什么知识图谱。用户提问可能在句首或句尾，请在这一步明确输出用户提问原句，在原句左右各用五个加号连接+++++； + 2. 详细分析用户目标，并按照用户目标分析图谱的实体分类、关系分类、每个实体的属性； + 3. 记录中间结果，列出实体列表，关系列表，每个实体的属性列表； + 4. 由上述结论总结得出用于指导大模型生成符合用户要求的Prompt，输出最终版本的提示词，该提示词需要包含第二步第三步中详细信息以及Json格式的示例，输出结果放置在下面括号内<<<<<>>>>> + 用户输入如下： + {UserInput} + """ + ##回复每个子问题，提取并标记用户要求，据此生成Prompt + result=API(GenerateTemplate4User) + df.loc[index,'分析过程']=result + + ##获取+标记的用户要求 + pattern1 = r'\+{3,}(.*?)\+{3,}' + matches_query=re.findall(pattern1,result,re.DOTALL) + ##过去模型生成的Prompt + pattern2 = r'<{3,}(.*?)>{3,}' + matches_template=re.findall(pattern2,result,re.DOTALL) + df.loc[index, '生成指令']=matches_template + + #测试模型生成的Prompt的建图效果 + BuildGraph4User=f"先完成：{matches_query[-1]}，再将输出知识图谱转化为Neo4j风格的可执行语句，用户输入如下{UserInput}" + result=API(BuildGraph4User) + df.loc[index,'应用该指令效果']=result + + df.to_excel(r'C:\Data_Test.xlsx', index=False) Review Comment: Consider moving the Excel write operation outside of the for loop to reduce repetitive disk writes and improve performance. ```suggestion # Write the updated DataFrame to Excel after processing all rows df.to_excel(r'C:\Data_Test.xlsx', index=False) df.to_excel(r'C:\Data_Test.xlsx', index=False) ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
