基于FastText模型的农业短文本分类研究

Classification of Agricultural Short Texts Based on FastText Model

本文提出基于FastText分类语言模型来解决农业短文本分类问题。在数据处理好类别的情况下,利用实验对12万条农业数据集进行实验。并探究与典型深度语言模型(TextRNN、TextCNN、TextDPCNN、Transformer)进行对比分析其中的分类准确率和分类处理响应时间。实验结果得出结论,基于深度学习的FastText模型的农业短文本分类效果最好,FastText模型对比其他模型的准确率、精确率、召回率和F1值提高了1%~4%。FastText模型可以对中文农业短文本分类处理速度更好,更优于其他典型深度语言模型算法。

This paper proposes a FastText classification language model to solve the problem of agricultural short text classification.Under the condition of data classification,experiments were carried out on 120,000 agricultural data sets.And compared with typical deep language models(TextRNN,TextCNN,TextDPCNN and Transformer),the classification accuracy rate and response time of classification processing are analyzed.The results show that the FastText model based on deep learning has the best effect on agricultural short text classification,and the FastText model improves the accuracy,accuracy,recall rate and F1 value by 1%~4%compared with other models.FastText model can classify Chinese agricultural short texts faster than other typical deep language model algorithms.