Review of Semantic Analysis Techniques of Agricultural Texts

Abstract

With the development of Internet and artificial intelligence technology, agricultural knowledge intelligent services have gradually assumed the role of providing effective technical guidance for agricultural production management, especially during the epidemic. The key technologies and applications in the semantic understanding of agricultural knowledge service texts were reviewed. Firstly, its progress in agriculture was introduced according to the semantic processing methods based on rules, machine learning and deep learning in natural language processing. Then, the semantic analysis method for the characteristics of agricultural knowledge was introduced, covering the storage, expression and calculation of the main process of agricultural text analysis, including knowledge extraction, knowledge fusion, knowledge representation and knowledge inference of agricultural knowledge graph. The representation model of agricultural text such as TF-IDF, Word2Vec and BERT and classification models such as CNN, RNN and Attention were presented. Then the common corpus was described. The application of semantic understanding in agriculture from the aspects of agricultural intelligent question answering, agricultural semantic retrieval and agricultural intelligent management decision as well were introduced. Finally, the research trend of agricultural text semantic understanding was prospected from the aspects of standardization construction of agricultural corpus, complexity of semantic understanding model, multi-modal semantic processing, multi-region and multi-language semantic understanding.

Keywords: agricultural knowledge intelligent service, deep learning, natural language processing, agricultral text, semantic analysis