Knowledge Management Forum

Construction of Intelligent Q&A System for Medicinal Plant Based on Multimodal Knowledge Graph

Doudou Zhao, Yujun Wang, Rui Liu, Chang Liu

Knowledge Management Forum, 2024, 9(5): 487-504. DOI: 10.13266/j.issn.2095-5472.2024.036

数据源名称	数据源内容
2015年版《中华人民共和国药典》一部	包括药材和饮片、植物油脂和提取物、成方制剂和单味制剂等
TCMID	TCMID包括配方、草药和草药成分；现代药理学和生物医学深入研究的药物和疾病的信息
	TCMID包括配方、草药和草药成分；现代药理学和生物医学深入研究的药物和疾病的信息
NPASS、CTD	NPASS提供了天然产物的物种来源和生物活性方面的详细信息 CTD整合大量化学物质、基因、功能表型和疾病之间相互作用数据
PPBC	PPBC收录各类植物图片共583科6 354属53 138种

Table 1 Data source

Other figure/table from this article

Figure 1 System architecture diagram
Figure 2 Multi-modal medicinal plant pattern layer
Table 2 Entity categories and quantities
Table 3 Relationship categories and quantities
Table 4 User question simulation data set
Table 5 Problem category classification effect
Table 6 Cypher s tatement query template
Table 7 Partial answer template
Figure 3 Training set accuracy curve
Figure 4 Validation set accuracy curve
Figure 5 Accuracy curves of different model training sets
Figure 6 Accuracy curves of different model validation sets
Table 8 Model accuracy comparison
Figure 7 Comparison of verification accuracy of two data sets
Figure 8 EfficientNet-B0-B4 training set accuracy curve
Figure 9 EfficientNet-B0-B4 validation set accuracy curve
Table 9 Efficiency comparison of EfficientNet series models
Table 10 Training accuracy and verification accuracy of EfficientNet-B3 model
Figure 10 System interface display
Figure 11 Example of medicinal plant text query
Figure 12 Example of medicinal plant image query
Table 11 Examples of questions provided by subjects
Figure 13 Examples of pictures provided by subjects