Research on the Chinese Named-Entity–Relation-Extraction Method for Crop Diseases Based on BERT

Round 1

Reviewer 1 Report

The article describes a novel entity-relation extraction model for crop diseases in Chinese languages, with the main component being a knowledge graph. The proposed model comprises of entity extraction and relation extraction steps. Importantly, the authors also describe a newly created Chinese crop disease corpus dataset.

- The research problem is well-motivated, as crop disease is a serious issue in China, and while there has been attempts for similar work in general agricultural domains, but not specifically in relation to crop diseases.

- I believe it would be beneficial for the paper to be slightly restructured in a way that general descirption of models (e.g. BERT, BiLSTM, etc.) should be presented first, and any specific details about the data used in this study, preprocessing etc. presented afterwards. It would mean for example that 2.1 is moved later, and any examples from the other subsections of Section 2 are integrated there. This would help the reader in seeing the actual contributions with appropriate examples in one place, separated from generic material.

- The results are convincing about the performance of the presented approach. In Tables 5 and 8, how are the measures calculated, is it the weighted/simple average of precision/recall/F1 for all entity types/relations? This could be explained more in the paper

- The authors could present more details about the knowledge graph, whether it was also evaluated by experts, offer some specific examples based on the shown graph, and how it can be used in practice. This could be of interest to readers who are not familiar with this component, as currently there is limited explanation and details in Section 3.1

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

In this manuscript, a Chinese named entity-relation extraction model for crop diseases (BBCPF) was proposed to integrate the fragmented textual data of crop disease knowledge and solve the current problems of disordered knowledge management, weak correlation, and knowledge sharing difficulties. The proposed BBCPF model is based on a knowledge graph that describes complex relationships among disease entities in a structured form. The proposed model consists of two parts, named entity recognition and relation extraction. To achieve better performance, the proposed model uses the BERT model, the BiLSTM layer, CRF, PERT and the fully connected layer. In addition, in this study a crop disease corpus dataset was created by searching agricultural technology websites and manually extracting data from reference books. To demonstrate the effectiveness of the proposed approach, experiments were performed on the relevant self-generated disease dataset. The results are presented for entity extraction and relation extraction, respectively. Furthermore, the proposed approach is compared with other recent relevant models and the obtained results are discussed. The authors have presented their work very well from both theoretical and practical point of view. The paper is technically sound and based on the fundamentals of the field, and the references given by the authors are applicable and sufficient (31 citations). However, the authors should add additional information in the references because it was difficult to verify some citations.

 

Please consider the following corrections and comments (P - Page, L - Line written in the left margin):

#) Regarding references:

Refence [3] Zhang, J.; Zhang, X.; Wu, C.; Zhao, Z., Survey of Knowledge Graph Construction Techniques, Computer Engineering, 2022, 48 (03), 23-37

When checking the references, I had a problem accessing the reference [3]. I received the response "The page is not working" from the host www.ecice06.com. Therefore, I would like to ask the authors if they can provide the DOI or another link so that the mentioned reference can be checked.

I was able to access the following, but it is not the citation listed at [3]: H. Zhou, T. Shen, X. Liu, Y. Zhang, P. Guo et al., "Survey of knowledge graph approaches and applications," Journal on Artificial Intelligence, vol. 2, no.2, pp. 89–101, 2020.

A similar problem with references [8] and [9]:

[8] Ruan, T.; Sun, C.; Wang, H.; Fang, Z.; Yin, Y., Construction of Traditional Chinese Medicine Konwledge Graph and Its Application, JOURNAL OF MEDICAL INFORMATICS 2016, 37 (04), 8-13.    

[9] Liu, Y.; Li, H., Survey on Domain Knowledge Graph Research, Computer Systems & Applications 2020, 29 (06), 1-12.

I suggest the authors to add additional information, perhaps a title in the source language, e.g. (in Chinese …). 

 

#) (P4, L126) Please provide references for “existing studies” in the sentence “Compared with the existing studies, the disease knowledge corpus established in this paper was supplemented with the labeling information of disease features, control methods, and pathogenic factors, so that the labeling granularity was much finer.”

#) Considering that the proposed model was created for Chinese domain knowledge of crop diseases, the authors should briefly discuss the possibilities and challenges of creating such a model, e.g., for English domain knowledge.

#) Please check proofreading and English spelling

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

My concerns are mostly resolved, although I'm still having trouble accessing references [3], but it may be a server availability issue.