Broad-Coverage German Sentiment Classification Model and Dataset for Dialog Systems

<strong>Training a Broad-Coverage German Sentiment Classification Model for Dialog Systems</strong> This paper describes the training of a general-purpose German sentiment classification model. Sentiment classification is an important aspect of general text analytics. Furthermore, it plays a vital role in dialogue systems and voice interfaces that depend on the ability of the system to pick up and understand emotional signals from user utterances. The presented study outlines how we have collected a new German sentiment corpus and then combined this corpus with existing resources to train a broad-coverage German sentiment model. The resulting data set contains 5.4 million labelled samples. We have used the data to train both, a simple convolutional and a transformer-based classification model and compared the results achieved on various training configurations. The model and the data set will be published along with this paper. You can find the code for training testing the models, that was published along with the paper in this repository. The <em>germansentiment</em> Python package contains a easy to use interface for the model that was published with this paper.**面向对话系统的高覆盖度德语情感分类模型训练** 本论文阐述了一款通用德语情感分类模型的训练过程。情感分类是通用文本分析的重要组成部分，同时在依赖系统捕捉与理解用户话语中情感信号能力的对话系统与语音交互界面中，同样发挥着至关重要的作用。本研究详述了如何采集全新的德语情感语料库，并将该语料库与现有资源相结合，以训练高覆盖度的德语情感分类模型。最终生成的数据集包含540万条标注样本。我们使用该数据集分别训练了简单卷积模型与基于Transformer的分类模型，并针对多种训练配置下的实验结果进行了对比分析。本模型与数据集将随本论文一同公开发布。本论文配套公开的模型训练与测试代码可在本仓库中获取。配套本论文发布的`germansentiment` Python库提供了该模型的简易调用接口。