深度学习是本世纪出现的新一代机器学习技术,深度学习技术的发展与应用对现代自然语言处理技术产生了深远的影响。本文讨论了自然语言处理技术在深度学习技术的推动下所取得的主要进展,以及近几年自然语言处理领域出现的新的技术产品和经典案例,特别分析并阐述了深度学习在文本词向量构建、磁性标注与命名实体识别相结合用于词义消歧、卷及神经网络文本自动分类、主题提取及文本内容相关性计算等关键自然语言处理任务中所发挥的重要作用,并介绍了词向量技术在水稻知识领域的作用、农业领域专有命名实体识别以及农业文献内容相关性计算等实际应用案例,并剖析了了相关技术实现细节。最后本文展望了今后一个时期自然语言处理技术的发展方向,以及其在农业领域的应用前景,并阐明了自然语言处理技术对农业领域智能化应用不可或缺的意义。

Deep learning is an emerging but rapidly advancing technology having a profound impact on modern natural language processing (NLP) technology. This paper discusses recent developments of NLP technology driven by deep neural networks (DNN), as well as new products and recent cases. In particular, the paper examines advances relevant to the agriculture domain, such as DNN-based word embedding vector construction, the computational ability to recognize and name domain-specific entities and agricultural literature terms. Additionally, it analyzes the implementation details of related technologies. Finally, the paper reviews the trends and outlook for NLP technology, highlighting the significance of NLP technology for intelligent applications in agriculture.