The Research on Construction Method of Government Procurement Intention Recognition Dataset
Panpan Chen, Xiaochuan Zhang, Xinlai Xing, Dafei Lin, Da Teng, Changmeng Yang · 2023
Intention detection can effectively extract key information from text, which helps to provide personalized services for government procurement users and improve user experiences. Currently, there is a lack of open datasets in the field of government procurement question-and-answer (Q&A) services. To address this issue, this paper first collected initial data of 2026 pieces and cleaned them. Secondly, to ease the subjectivity brought by manual classification, text clustering algorithms were introduced to group the textual data. Finally, the data in each cluster was checked for rationality and labeled for intention through manual verification. K-Folder cross validation experiments were performed on the obtained dataset using a classification model, and the results showed that the dataset can be used in the practical application of dialogue systems and intention recognition in related fields. It also helps to improve service quality and user experiences more specifically in the context of government procurement Q&A.