{"id":13535261,"url":"https://github.com/WenRichard/KBQA-BERT","last_synced_at":"2025-04-02T00:33:08.221Z","repository":{"id":41377486,"uuid":"182223042","full_name":"WenRichard/KBQA-BERT","owner":"WenRichard","description":"基于知识图谱的问答系统，BERT做命名实体识别和句子相似度，分为online和outline模式","archived":false,"fork":false,"pushed_at":"2021-12-16T12:05:22.000Z","size":1549,"stargazers_count":1449,"open_issues_count":21,"forks_count":349,"subscribers_count":17,"default_branch":"master","last_synced_at":"2024-11-02T23:32:46.183Z","etag":null,"topics":["bert","knowledge-graph","nlpcc2017","sentence-similarity"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/WenRichard.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2019-04-19T07:36:28.000Z","updated_at":"2024-10-08T02:26:56.000Z","dependencies_parsed_at":"2022-08-10T02:06:59.377Z","dependency_job_id":null,"html_url":"https://github.com/WenRichard/KBQA-BERT","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WenRichard%2FKBQA-BERT","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WenRichard%2FKBQA-BERT/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WenRichard%2FKBQA-BERT/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WenRichard%2FKBQA-BERT/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/WenRichard","download_url":"https://codeload.github.com/WenRichard/KBQA-BERT/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":246735358,"owners_count":20825222,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["bert","knowledge-graph","nlpcc2017","sentence-similarity"],"created_at":"2024-08-01T08:00:52.242Z","updated_at":"2025-04-02T00:33:07.603Z","avatar_url":"https://github.com/WenRichard.png","language":"Python","funding_links":[],"categories":["BERT  Knowledge Graph Task :","知识图谱问答KBQA、多跳推理","Python","Open source projects"],"sub_categories":["其他_文本生成、文本对话","Others"],"readme":"# KBQA-BERT\n## 基于知识图谱的问答系统，BERT做命名实体识别和句子相似度，分为online和outline模式\n\n## Introduction\n本项目主要由两个重要的点组成，一是**基于BERT的命名实体识别**，二是**基于BERT的句子相似度计算**，本项目将这两个模块进行融合，构建基于BERT的KBQA问答系统，在命名实体识别上分为online predict和outline predict；在句子相似度上，也分为online predict和outline predict，2个模块互不干扰，做到了高内聚低耦合的效果，最后的kbqa相当于融合这2个模块进行outline predict，具体介绍请见[我的知乎专栏](https://zhuanlan.zhihu.com/p/62946533)！\n\n### ------------------------------------------- 2019/6/15 更新 ----------------------------------------  \n**把过去一段时间同学们遇到的主要问题汇总一下，下面是一些FAQ：**  \n  \n**Q:** 运行run_ner.py时未找到dev.txt,请问这个文件是怎么生成的呢？  \n**A:** 这一部分我记得当初是没有足够多的数据，我把生成的test.txt copy, 改成dev.txt了。  \n  \n**Q:** 你好，我下载了你的项目，但在运行run_ner的时候总是会卡在Saving checkpoint 0 to....这里，请问是什么原因呢？  \n**A:** ner部分是存在一些问题，我也没有解决，但是我没有遇到这种情况。微调bert大概需要12GB左右的显存，大家可以把batch_size和max_length调小一点，说不定会解决这个问题！。  \n  \n**Q:** 该项目有没有相应的论文呢？  \n**A:** 回答是肯定的，有的，送上 [**论文传送门**!](http://www.cnki.com.cn/Article/CJFDTotal-DLXZ201705041.htm)  \n\n**Q:** 数据下载失败，不满足现有数据？  \n**A:** 数据在Data中，更多的数据在[**NLPCC2016**](http://tcci.ccf.org.cn/conference/2016/pages/page05_evadata.html) 和[**NLPCC2017**](http://tcci.ccf.org.cn/conference/2017/taskdata.php)。    \n\n**PS：这个项目有很多需要提高的地方，如果大家有好点子，欢迎pull，感谢！这段时间发论文找工作比较忙，邮件和issue没有及时回复望见谅！**\n### ------------------------------------------- 2019/6/15 更新 ----------------------------------------  \n### 环境配置\n\n    Python版本为3.6\n    tensorflow版本为1.13\n    XAMPP版本为3.3.2\n    Navicat Premium12\n    \n### 目录说明\n\n    bert文件夹是google官方下载的\n    Data文件夹存放原始数据和处理好的数据\n        construct_dataset.py  生成NER_Data的数据\n        construct_dataset_attribute.py  生成Sim_Data的数据\n        triple_clean.py  生成三元组数据\n        load_dbdata.py  将数据导入mysql db\n    ModelParams文件夹需要下载BERT的中文配置文件：chinese_L-12_H-768_A-12\n    Output文件夹存放输出的数据\n    \n    基于BERT的命名实体识别模块\n    - lstm_crf_layer.py\n    - run_ner.py\n    - tf_metrics.py\n    - conlleval.py\n    - conlleval.pl\n    - run_ner.sh\n    \n    基于BERT的句子相似度计算模块\n    - args.py\n    - run_similarity.py\n    \n    KBQA模块\n    - terminal_predict.py\n    - terminal_ner.sh\n    - kbqa_test.py\n    \n ### 使用说明\n    \n    - run_ner.sh\n    NER训练和调参\n    \n    - terminal_ner.sh\n    do_predict_online=True  NER线上预测\n    do_predict_outline=True  NER线下预测\n    \n    - args.py\n    train = True  预训练模型\n    test = True  SIM线上测试\n    \n    - run_similarity.py\n    python run一下就可以啦\n    \n    - kbqa_test.py\n    基于KB的问答测试\n  \n ### 实验分析  \n ![NER图]( https://github.com/WenRichard/KBQA-BERT/raw/master/image/NER.jpg \"分析图\") \n \n ![kb图]( https://github.com/WenRichard/KBQA-BERT/raw/master/image/KB.png \"分析图\") \n \n **Cite**     \n如果你在研究中使用了KBQA-BERT，请按如下格式引用：  \n\n```\n@software{KBQA-BERT,\n  author = {ZhengWen Xie},\n  title = {KBQA-BERT: Knowledge Base Question Answering based BERT},\n  year = {2019},\n  url = {https://github.com/WenRichard/KBQA-BERT},\n}\n```\n \n --------------------------------------------------------------\n**如果觉得我的工作对您有帮助，请不要吝啬右上角的小星星哦！欢迎Fork和Star！也欢迎一起建设这个项目！**    \n**有时间就会更新问答相关项目，有兴趣的同学可以follow一下**  \n**留言请在Issues或者email  richardxie1205@gmail.com**\n    \n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FWenRichard%2FKBQA-BERT","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FWenRichard%2FKBQA-BERT","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FWenRichard%2FKBQA-BERT/lists"}