Abstract

Happiness becomes a rising topic that we all care about recently. It can be described in various forms. For the text content, it is an interesting subject that we can do research on happiness by utilizing natural language processing (NLP) methods. As an abstract and complicated emotion, there is no common criterion to measure and describe happiness. Therefore, researchers are creating different models to study and measure happiness.
最近，幸福成为了一个我们都关心的话题。它可以用各种形式来描述。对于文本内容，利用自然语言处理(NLP)方法对幸福感进行研究是一个有趣的课题。幸福作为一种抽象而复杂的情感，没有统一的标准来衡量和描述。因此，研究人员正在创建不同的模型来研究和衡量幸福。

In this paper, we present a deep-learning based model called Senti-BSAS (BERT embedded bi-lstm with Self-Attention mechanism along with the Sentiment computing). Given a sentence that describes how a person felt happiness recently, the model can classify the happiness scenario in the sentence with two topics: was it controlled by the author (label ‘agency’), and was it involving other people (label ‘social’).
在本文中，我们提出了一种基于深度学习的模型，称为Senti-BSAS(带有自注意机制和情感计算的BERT嵌入式Bi-LSTM)。给出一个描述一个人最近感觉幸福的句子，该模型可以将句子中的幸福场景分为两个主题:它是由作者控制的吗(标记为“代理”)，以及它是否涉及到其他人(标记为“社交”)。

Besides language models, we employ the label information through sentiment computing based on lexicon. The model performs with a high accuracy on both ‘agency’ and ‘social’ labels, and we also make comparisons with several popular embedding models like Elmo, GPT. Depending on our work, we can study the happiness at a more fine-grained level.
除了语言模型外，我们还通过基于词典的情感计算来利用标签信息。该模型在“代理”和“社会”标签上都有很高的准确性，我们还与几个流行的嵌入模型进行了比较，如Elmo, GPT。根据我们的工作，我们可以在更细粒度的水平上研究幸福。

Conclusions

In this paper we present a binary classification model: Senti-BSAS, which can classify your happiness is or is not controlled by yourself and is or is not involving other people at a high accuracy. Using BERT can make the word embedding part much better, Bi-LSTM performs better than usual LSTM model on contextual information. Self-attention mechanism also makes the result better. We also utilize the label information by sentiment computing, as we do the experiment on the VAD lexicon and find out that VAD value is related to the ‘agency’ and ‘social’ label.
在本文中，我们提出了一个二元分类模型：Senti-BSAS，它可以对你的幸福是否由自己控制以及是否涉及他人进行高精度的分类。使用
BERT模型可以使词嵌入部分得到更好的表现，Bi-LSTM模型在上下文信息上比一般的LSTM模型有更好的表现。自我注意机制也使结果更好。我们还通过情感计算来利用标签信息，在VAD词典上进行实验，发现VAD值与“代理”和“社会”标签相关。

Happiness is kind of an emotional concept, we can hardly measure it, therefore the study on the happiness is hard to go further. However, we can do it step by step, as we can clarify several huge concepts, then come into smaller ones. “Agency” and “social”, as two novel concepts, will lead to other happiness’ problem: after classifying agency label, we attain happiness moment which author made it happen, then we can further study how to make ourselves feel happiness, and as to social label, we can further find out how other people can affect with our happiness, what kind of people make you feel happiness and so on.
幸福是一种情感概念，我们很难衡量它，因此对幸福的研究也很难深入。然而，我们可以一步一步地做，因为我们可以先澄清几个大概念，然后再细化成较小的概念。能动性和社会性，作为两个新概念，会导致其他的幸福问题:机构分类标签后,我们获得幸福的时刻,作者使它发生,那么我们就可以进一步研究如何让自己感到快乐,和社会标签,我们可以进一步发现其他人如何影响与我们的幸福,什么样的人让你感觉幸福等等。

标签：BERT,Senti,based,幸福,模型,label,LSTM,happiness
来源： https://blog.csdn.net/qq_33790600/article/details/120891423