QA Survey CN

QA-Survey

Question and answer systems are a research direction in the fields of artificial intelligence and natural language processing that has attracted much attention and has broad development prospects. The main reason for its rise in research is people's need to quickly and accurately obtain information. During the research and development of the Q&A system, the research team of Beihang Big Data High-tech Center focused on the research of the Knowledge Graph-based Q&A (KBQA), the Text-based Q&A (TextQA), the Table-based Q&A system (TableQA) and the Vision-based Q&A system (VisualQA), and integrated its existing academic and industrial related theories and technologies, currently including the following eight parts:

• Community Question Answering (CQA)-Academic Circle
• Community Question Answering (CQA)-Industrial
• Machine Reading Comprehension (MRC)-Academic
• Machine Reading Comprehension (MRC)-Industrial
• Knowledge Base Question Answering (KBQA)-Academic Circle
• Knowledge Base Question Answering (KBQA)-Industrial
• Table Question Answering (TQA)-Academic Circle
• Visual Question Answering (VQA)-Academic Circle

Hope to provide assistance to relevant scholars and researchers in the fields of question-answer systems and natural language processing. This survey will maintain regular and continuous updates and continuously track cutting-edge technologies. If there are any shortcomings, please criticize and correct them. All Q&A systems and natural language processing researchers are welcome to use them, and everyone is also welcome to improve this survey together.

News!

• CCL2022 evaluation: knowledge extraction evaluation for domain-oriented Q&A. The assessment was jointly organized by our team and Aviation Travel Zongheng. Everyone is welcome to pay attention and participate!
• Pre-trained Chinese reading comprehension model for question-and-answer and extraction applications. We have sorted out and obtained data in the Chinese field, and based on optimizing relevant models, we have achieved leading results in reading comprehension in the Chinese field. (20220630)
• Cross-domain knowledge intelligent interaction platform embedded map question and answer algorithm, supporting the rapid configuration of knowledge graphs in different fields to launch the Q&A application platform (20220625)
• The recent research progress of visual Q&A in ACL2021, ACL2022, EMNLP2021, NAACL2021, NAACL2022, CVPR2021, and ICCV2021 will be updated soon. (It is expected to be updated in mid-July 2022)
• The related work of knowledge graph Q&A in ACL2022, NAACL2022, AAAI2022 and SIGIR2022 has been updated. (20220618)
• Adding Logical Reasoning Logical Reasoning Reading Comprehension Related Datasets and Models (20220612)
• Reading comprehension related work in ACL2022, AAAI2022 and SIGIR2022 has been updated. (20220612)

Open source contribution

We are currently preparing the English version of Survey. We hope that the majority of researchers can participate, so that the survey will continue to improve and evolve. We will also provide promotion platforms and corresponding material incentives for participants. If you are interested in participating in open source contributors, please contact [email protected] at the email address. The email will be replied within 3 working days! I hope that this survey can further contribute to QA research and application in everyone's efforts.

Related results

Academic Papers

• Richong Zhang, Yue Wang, Yongyi Mao and Jinpeng Huai: Question Answering in Knowledge Bases: A Verification Assisted Model with Iterative Training. ACM Transactions on Information Systems. 37(4): 40:1-40:26 (2019)

• Yue Wang, Richong Zhang, Cheng Xu and Yongyi Mao: The APVA-TURBO Approach To Question Answering in Knowledge Base. Proceedings of the 27th International Conference on Computational Linguistics(COLING), pages 1998–2009 (2018)

books

• It will be officially released in the second half of 2022!

patent

• CN107748757A | A question-and-answer method based on knowledge graph
• CN111858887A | A community question and answer system for airport service
• CN111858888A | Multi-round dialogue method for value airport scene
• Knowledge graph construction method based on multi-source data
• Design and implementation of Q&A system based on knowledge graph
• Research on Q&A methods based on knowledge base

Application Products

• Intelligent Q&A on Knowledge and Action Knowledge Graph
• General domain text reading comprehension
• Cross-domain knowledge intelligent interaction platform
• Civil Aviation Encyclopedia Knowledge Q&A
• Vertical field Q&A application practice

Acknowledgements

We hereby thank Beihang High-tech Center, which supports data disclosure and system research and development, and all team members involved in this work:

Nie Zhijie, Hu Zhiyuan, Zhang Cunwang, Huang Hailang, Li Hangyu, Zhang Shuhui, Zhang Yanzhao, Li Yitong, Liu Chumin, Tian Yuan, Wang Junkai, Zhang Minghui

about Us

Beijing Big Data Science and Brain Computer Intelligence High-Primary Innovation Center