基于图像和LM的标准术语检测技术比较研究

Comparativeresearch on standard term detection techniques based on image and LM

标准术语是指特定领域内具有特定含义和用途的字词组合。在使用标准术语时,应确保没有缺字、添字、替换字或字序错乱等情况,特别是在涉及意识形态领域时更应如此。文章回顾了图像处理、预训练语言模型等深度学习技术的发展历程,并探讨了深度学习技术在标准术语检测中的应用;提出了基于深度学习图像特征和基于预训练语言模型的2种方法来实现标准术语的检测,并对自行构建的测试数据进行了验证。实验结果显示,基于预训练语言模型的标准术语检测方法表现更优,平均准确率达到了99.4%。文章采用的方法适用于各个领域,尤其在意识形态领域具有广泛应用价值。

Standard terminology refers to a combination of words with specific meanings and uses within a specific field.When using standard terminology,it is important to ensure that there are no missing words,additions,substitutions,or disordered word order,especially when it comes to ideological fields.The article reviews the development process of deep learning techniques such as image processing and pre trained language models,and explores the application of deep learning techniques in standard terminology detection.Two methods based on deep learning image features and pre trained language models were proposed to achieve standard terminology detection,and self constructed test data was validated.The experimental results show that the standard term detection method based on pre trained language models performs better,with an average accuracy of 99.4%.The method adopted in the article is applicable to various fields,especially in the field of ideology,which has broad application value.