当前位置: X-MOL 学术J. Circuits Syst. Comput. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Automatic Question Answering System Based on Convolutional Neural Network and Its Application to Waste Collection System
Journal of Circuits, Systems and Computers ( IF 0.9 ) Pub Date : 2020-06-30 , DOI: 10.1142/s0218126621500134
Chuan Jiang 1 , Qianmin Su 1 , Lele Zhang 1 , Bo Huang 1
Affiliation  

As a typical cyber-physical-social system (CPSS), the waste collection system profoundly changes the current waste processing mode and greatly relieves the dilemma of waste disposal. However, the existing waste collection system does not provide the function that guides people to deliver the waste into the correct trash bin. In order to improve the efficiency of waste collection system, we propose an automatic question answering system based on convolutional neural network (CNN) to help people classify waste correctly. The construction process of automatic question answering system is divided into the following steps. We first construct a question answering dataset about waste classification, in which question answering pairs from the four waste categories (recyclable waste, harmful waste, dry waste, and wet waste) are included. After the dataset is constructed, we perform text preprocessing on the dataset, which includes denoising, Chinese word segmentation, and removing stop words. After text preprocessing, we use the Word2vec model as feature representation. Then, we construct a CNN and utilize the word embeddings as an input to train model. Finally, we deploy the trained model to the waste collection system, which can answer the question of waste classification that people ask. We also present a comparative analysis of the proposed method and traditional machine learning methods. The experiment shows that the proposed method has higher accuracy of waste classification than that of traditional machine learning methods.

中文翻译:

基于卷积神经网络的自动问答系统及其在垃圾收集系统中的应用

作为典型的网络-物理-社会系统(CPSS),垃圾收集系统深刻改变了当前的垃圾处理模式,极大地缓解了垃圾处理的困境。然而,现有的垃圾收集系统并没有提供引导人们将垃圾送入正确垃圾桶的功能。为了提高垃圾收集系统的效率,我们提出了一种基于卷积神经网络(CNN)的自动问答系统,以帮助人们正确分类垃圾。自动问答系统的构建过程分为以下几个步骤。我们首先构建了一个关于垃圾分类的问答数据集,其中包括来自四个垃圾类别(可回收垃圾、有害垃圾、干垃圾和湿垃圾)的问答对。数据集构建完成后,我们对数据集进行文本预处理,包括去噪、中文分词和去除停用词。在文本预处理之后,我们使用 Word2vec 模型作为特征表示。然后,我们构建一个 CNN 并利用词嵌入作为训练模型的输入。最后,我们将训练好的模型部署到垃圾收集系统,可以回答人们提出的垃圾分类问题。我们还对所提出的方法和传统的机器学习方法进行了比较分析。实验表明,所提方法比传统机器学习方法具有更高的垃圾分类准确率。在文本预处理之后,我们使用 Word2vec 模型作为特征表示。然后,我们构建一个 CNN 并利用词嵌入作为训练模型的输入。最后,我们将训练好的模型部署到垃圾收集系统,可以回答人们提出的垃圾分类问题。我们还对所提出的方法和传统的机器学习方法进行了比较分析。实验表明,所提方法比传统机器学习方法具有更高的垃圾分类准确率。在文本预处理之后,我们使用 Word2vec 模型作为特征表示。然后,我们构建一个 CNN 并利用词嵌入作为训练模型的输入。最后,我们将训练好的模型部署到垃圾收集系统,可以回答人们提出的垃圾分类问题。我们还对所提出的方法和传统的机器学习方法进行了比较分析。实验表明,所提方法比传统机器学习方法具有更高的垃圾分类准确率。我们还对所提出的方法和传统的机器学习方法进行了比较分析。实验表明,所提方法比传统机器学习方法具有更高的垃圾分类准确率。我们还对所提出的方法和传统的机器学习方法进行了比较分析。实验表明,所提方法比传统机器学习方法具有更高的垃圾分类准确率。
更新日期:2020-06-30
down
wechat
bug