Year: 2023
Type: conference-paper
Abstract: The lack of large-scale information operation and maintenance corpus greatly limits the development of information technology operation and maintenance management, especially in non-English languages. To improve this situation, in this paper, we introduce a large-scale Chinese information operation and maintenance knowledge retrieval corpus and release it publicly. How to collect a large amount of retrieval corpora in different languages is a key point in building such a corpus. In this paper, we first collect a large amount of Chinese information operation and maintenance knowledge corpus related to high-frequency words in various fields using search engines, and then generate relevant questions for the corpus using ChatGPT (https://chat.openai.com/). Finally, we recruit three annotators to manually check the quality of the retrieval corpus. After this process, we have built a Chinese information operation and maintenance knowledge corpus containing 2000 retrieval questions. To verify the quality of the corpus, we divide it into two parts: a training set containing 1500 retrieval questions and a test set containing 500 retrieval questions, and test several well-known retrieval methods on them (https://pan.baidu.com/s/1rLWqHZJhE9nEOYg3OTC1Ag). The experimental results not only prove the high quality of the corpus but also provide a solid baseline performance for further research on this corpus.
Authors: Kai Sun, Tiancheng Zhao, Yangling Chen, Shuo Han, Yue Wu, Nan Xiang
Institution: State Grid Nanjing Power Supply Company (China)
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4396648474