Implicit Emotion Analysis Based on Improved Supervised Contrastive Learning

A dual framework for implicit and explicit emotion recognition: An ensemble of language models and computational linguistics

The regulatory effect of embodied emotion on one’s general emotion and the impact of the compatibility or incompatibility of the two types of emotion on creative thinking are still debatable. The purpose of this study is to investigate these issues experimentally. In Experiment 1, participants completed an explicit positive and negative emotion test [Positive and Negative Affect Schedule (PANAS)] and an implicit positive and negative emotion test [Implicit Positive and Negative Affect Test (IPANAT)] twice on a computer after emotional video priming was used to induce negative emotions and facial expression manipulation was performed to induce embodied positive or negative emotions. It was found that maintaining the expression of a suppressed smile was helpful in regulating negative emotions (p = 0.047). Specifically, the implicit negative emotions induced by facial expression manipulation had a positive regulating effect on the implicit negative emotions induced by the video (T1, M = 47.813; to T2, M = 44.188). In Experiment 2, the positive or negative emotions of the participants were induced using emotional videos, and facial expression manipulation was used to induce their embodied positive or negative emotions. Then, the participants completed a creative test by completing alternative use tasks (AUTs) and Chinese character riddles. The AUT fluency score in the emotionally compatible group was significantly higher than that in the emotionally incompatible group (p = 0.032), but while experiencing negative emotions, the emotionally compatible group had a significantly higher originality score and insight in Chinese character riddle score than the emotionally incompatible group (p = 0.017, p = 0.004). Therefore, embodied negative emotion has a significant regulating effect on implicit negative emotion. The compatibility of emotion activated by facial expression and viewing a video contributes to creative thinking, whereas the incompatibility of emotion hinders creative thinking. The compatibility of emotion under positive emotions improved thinking fluency, whereas under negative emotions, it activated originality and insight in creative thinking. The influence of such emotional compatibility on creative thinking may be due to the regulating effect of embodied emotions on implicit emotions induced by emotional stimuli.

Implicit emotions are often expressed through implicit and weak clues between modalities due to the lack of explicit emotional feature words, representing a significant challenge for multimodal sentiment analysis. In order to improve implicit emotion recognition, this paper proposes a multimodal sentiment analysis method that integrates KAN and the modal dynamic fusion mechanism. This method first introduces the KAN structure to construct a modal feature encoder to enhance the emotional expression ability of features. Then, the emotional contribution weight of each modality is calculated using the difference between the unimodal and multimodal sentiment scores, and the cross-attention mechanism guided by the main modality is used for feature fusion. Experiments on four datasets, CH-SIMS, CH-SIMSv2, MOSI, and MOSEI, show that the proposed method significantly outperforms the mainstream model in multiple indicators, especially when dealing with samples with implicit or ambiguous emotional expressions. The results verify the effectiveness of enhancing feature encoding capabilities and utilizing modal asymmetry information in implicit sentiment analysis.

In recent years, vocabulary emotion processing has become immensely popular and the requirements for language emotion analysis mining and processing have become significantly abundant. The sentiment extraction and analysis work has always been very challenging; especially, the Chinese word segmentation operation is difficult to deal with effectively, the multiple combinations of implicit and explicit words make the task of sentiment analysis mining more difficult, and, in particular, the efficiency of machine analysis of language sentiment is feeble. We use some expressions and sentiment vocabulary dictionaries combined with hybrid structures and use information synergy methods to get in touch with sentiment analysis methods. We use the relevant sentiment to evaluate the explicit or implicit emotional association of the emotional connection of the vocabulary and add the unique emotional word matrix to analyze the related clustering results of the emotional words to continuously optimize and upgrade the performance, so that our sentiment analysis results are systematic in terms of efficiency and significantly improved.

Aiming at the problem that normal vector representation method cannot fully represent the emotional semantic information contained in the text when dealing with Chinese text sentiment analysis task, a multi-granularity convolutional capsule neural network model is constructed. The input of sentiment analysis model is vector representation of the texts trained through the language model. As there exists the problem that a single language model is not enough to abstract text features, a multi-granularity text vector representation method based on BERT is proposed. The text vector representations with different granularities are input into the improved multi-channel convolutional capsule neural network. The capsule layer can associate the low-level and high-level text features, and it will extract information selectively through the dynamic routing algorithm to construct emotional and semantic features of the whole text. Multiple comparative experiments confirm that the method proposed in this paper will efficiently improve the accuracy of Chinese sentiment analysis.

In recent years, there has been significant progress in Chinese text sentiment analysis research. However, few studies have investigated the differences between languages, the effectiveness of domain knowledge, and the requirements of downstream tasks. Considering the uniqueness of Chinese text and the practical needs of sentiment analysis, this study aims to address these gaps. To achieve this, we propose a method that deeply integrates the knowledge vector obtained from the emotional knowledge triplets using the TransE model with feature vectors from models like BiGRU and attention mechanisms. We also introduce radical features and emotional part of speech features based on the characteristics of characters and words. In addition, we propose a collaborative approach that integrates characters, words, radicals, and multi-granularity semantic features such as part of speech. Our approach, as evidenced by the Douban Film Review dataset and the NLPECC dataset, proficiently leverages emotional insights alongside nuanced linguistic elements, significantly bolstering the accuracy of sentiment detection in Chinese texts. The method achieved F1-score of 89.23% and 84.84%, respectively, underscoring its efficacy in the realm of Chinese sentiment analysis.

Sentiment analysis of Chinese micro-blog based on sentiment dictionary has become a challenging research subject in the field of artificial intelligence. However, due to insufficient sentiment words, Chinese micro-blog sentiment analysis is difficult to process high accuracy. Aimed at this issue, we propose a method for constructing multiple sentiment dictionaries, which mainly constructs original sentiment dictionary, emoji dictionary, and other related dictionaries. Among them, we have innovatively constructed a Chinese micro-blog new word sentiment dictionary. Multiple sentiment dictionaries increase the coverage of sentiment words. At the same time, we further analyze semantic rule sets between Chinese micro-blog texts and take the inter-sentence analysis rules and sentence pattern analysis rules into the sentiment analysis of Chinese micro-blog, which further improves the accuracy of Chinese micro-blog sentiment analysis. Finally, based on the method of multiple sentiment dictionaries and semantic rule sets, we propose an algorithm for Chinese micro-blog sentiment calculation from complex sentences to clauses and then from clauses to words, and finally combined with the emoji. This algorithm can accurately classify Chinese micro-blog into positive Chinese micro-blog, negative Chinese micro-blog, and neutral Chinese micro-blog. The experimental results show that this method has greatly improved the sentiment analysis of Chinese micro-blog.

Text sentiment analysis technology is a hot topic recently. As a short text, there is a feature of using complex sentences to express the author's true views and complex emotional tendencies in micro-blogs. In current researches on sentiment classification based on machine learning, few of them focus on complex sentences. This paper proposed a sentiment analysis method based on SVM and complex phrasing classifier, and made a full analysis of structural features of Chinese conditional sentences, transition sentences and multiple negative sentences, which were taken as text features. A variety of different combinations of features were chosen, including emotional words, speech, negative words, the degree of adverbs and punctuation, etc., to optimize the results of sentiment analysis through multiple sets of experiments. The experiments show that when we choose the combinations of features of emotional words, part of speech and complex sentence patterns, this method improved the accuracy of sentiment classification compared to the common method.

Many theories of driver behaviour suggest that unconscious or implicit emotions play a functional role in the shaping and control of behaviour. This has not been experimentally tested however. Therefore, in this study the effects of emotive masked images on driver behaviour were examined. While driving a simulator, participants were repeatedly exposed to negative or neutral emotionally laden target images that were sandwich masked by emotionally neutral images. These images were encountered across two different trials each of which consisted of 3–4 minutes of driving on a rural road. The results indicate an effect of the negative target images primarily in reducing the extent of familiarisation occurring between the first and second experimental drives. This is evident in a reduced decrease in heart rate and a reduced increase in high band heart rate variability and actual travelling speed from the first to second drives if the negative target image was presented in the second drive. In addition to these findings there was no clear effect of the target image on subjective ratings of effort or feelings of risk. There was however an effect of gender, with the majority of the effects found in the study being limited to the larger female dataset. These findings suggest that unconscious or implicit emotional stimuli may well influence driver behaviour without explicit awareness.

The long text in our study involves a complete discourse structure with a title and approximately 1500 Chinese characters within. Long text sentiment analysis has faced a lot of difficulties on the grounds that a long text often comprises multiple sentiments and diverse focuses. This paper proposes to analyze the sentiment of long texts based on the combination of title and topic sentence. Based on the fine-labeled texts, the analysis extracted the important features of topic sentences, such as location, feature words, degree of topic relevance and emotional words. Then two topic sentences were extracted from each text that can best represent the topic through weighted calculation of these multi-dimensional features. Next this paper combined the two topic sentences with the title to complete the sentiment analysis of the whole document. The topic sentence extraction method has achieved good results in the task of extracting and judging key emotional sentences of news in the 6th COAE (Chinese Opinion Analysis Evaluation) (2014), which shows that the method is effective. This method has been put into practice in the National Language Public Opinion Monitoring system, with an accuracy of 0.82.

Homecoming, more traditionally known as Mudik, has become a trending topic on several social media platforms as soon as the 11-day homecoming ritual ban was announced on 7 April 2021. Opinions, varying from those in favor of and against the ban, start to rapidly appear. Twitter, a social media platform which is now considered to be an extension of oneself and often used to express ones’ opinion, has become flooded with comments on the homecoming ritual ban. The swarm of opinions in the form of tweets were then used as a dataset for sentiment analysis in order to understand how people perceive the ban. The algorithm used in this research is the classification algorithm using the Support Vector Machine method. The dataset was classified into three sentiments: positive, negative, and neutral. The use of the Support Vector Machine algorithm yielded a 62% accuracy with this dataset. The sentiment analysis showed that the keyword "mudik" had a neutral sentiment for the most part. Meanwhile, results of engagement analysis show that the largest forms of engagements were retweets and liking tweets that had a neutral sentiment. When the neutral sentiment was removed, we found that the largest sentiment on the homecoming ritual ban was negative. This is likely due to the release of an addendum to the Covid-19 Handling Task Force Circular Number 13 of 2021 on 22 April 2021 that imposes more restrictions on and extends the effective dates of the restrictions related to the homecoming ritual ban; exactly one day before the data scraping of 5000 datasets on tweets from 23 April 2021 was carried out. The researcher had already sampled the tweets with the most engagements (those with the most retweets and likes). It was found that some tweets had a negative sentiment, but the model classified it as having a neutral sentiment. This may be affected by inaccuracies of dataset training as some of the tweets were in Malay rather than Indonesian. A challenge that needs to be overcome is the limited number of datasets for NLP training or sentiment analysis for the Indonesian language in comparison to that of the English language. On the other hand, this has become an opportunity for the researcher to develop a more appropriate training model.

Implicit sentiment analysis is an important part of sentiment computing, especially sentiment analysis based on deep learning has become a research hotspot in recent years. In this paper, convolutional neural networks are used to extract text features, combined with long-term and short-term memory network (LSTM) structure to extract context information, and add attention mechanism to the network to construct a new hybrid neural network model to achieve implicit emotion for text analysis. The hybrid neural network model extracts more meaningful sentence semantic and structural hidden features from the word-level and sentence-level hierarchies, respectively, and pays attention to the features with a large emotional contribution rate through the attention mechanism. The classification accuracy rate of the model on the public implicit sentiment dataset has reached 77%. The research of implicit sentiment analysis can improve the effect of text sentiment analysis more comprehensively, and further promote the application of text sentiment analysis in the fields of knowledge embedding, text representation learning, user modeling and natural language.

Topic-related Chinese message sentiment analysis

In the target article, Sabini and Silver (this issue) make the important point that emotional words and emotional experiences do not have a one-to-one map ping. In making this assertion, they join a long list of re searchers and theorists including Kagan (1988), Lang (1993) and others who have reached the same conclu sions albeit for quite different reasons. Sabini and Silver assert that there are fewer emotional experiences than there are emotional words. Kagan has claimed essen tially the opposite. Kagan noted that psychologists are wont to use basic emotion labels such as fear to refer to a very wide range of phenomena. To continue with this ex ample, the state experienced by a rodent in response to a simple tone that has been paired with shock is called fear. That same word is used to describe the state expe rienced by a person as she enters a doctor's office to learn the outcome of a breast biopsy. Although, there may in deed be some superficial resemblance between these states, it is also clear that they differ in important ways that are likely to involve different underlying neurobiological systems. In fact, when direct measures of brain function are examined, such different kinds of fear are associated with different patterns of brain activ ity (see e.g., Davidson & Irwin, 1999 for review), yet psychologists often use the same word to refer to these very different states. Even though reaching the same conclusion as Sabini and Silver, Kagan and others (e.g., Davidson, 1993) believe that there are far fewer emotion words than there are emotion experiences and the use of the same word to refer to very different states that are elicited in vastly different contexts is hazardous and po tentially misleading. Lang's perspective on emotion offers a different view on why emotion words and emotion experiences often diverge. Lang (1993) has suggested that there are three major domains that contribute to emotion-ac tion, physiology, and subjective experience. Within each of these domains, many further distinctions can be made but just among these three domains, Lang and his colleagues have argued that there are imperfect as sociations at best, since each reflects different features of emotion that do not necessarily cohere (Bradley & Lang, 2000; Lang, Levin, Miller, & Kozak, 1983). For instance, Lang et al. (1983) observed the relative inde pendence among verbal report of affect, overt behavior and psychophysiological measures in response to cer tain commonly used methods of emotion induction. Bradley and Lang (2000) report that in response to mild to moderate elicitors of emotion, covariation be tween the response systems explains 15% of the vari

Chinese short text sentiment analysis can help understand society’s views on various hot topics. Many existing sentiment analysis methods are based on sentiment dictionaries. Still, sentiment dictionaries are easily affected by subjective factors. They require a lot of time to build as well as maintenance to prevent obsolescence. For the aim of extracting rich information within texts more effectively, we propose a Channel Attention TextCNN with Feature Word Extraction model (CAT-FWE). The feature word extraction module helps us choose words that affect the sentiment of reviews. Then, these words are integrated with multi-level semantic information to enhance the information of sentences. In addition, the channel attention textCNN module that is a promotion of traditional TextCNN tends to pay more attention to those meaningful features. It eliminates the impacts of features that do not make any sense effectively. We apply our CAT-FWE model to both fine-grained classification and binary classification tasks for Chinese short texts. Experiment results show that it can improve the performance of emotion recognition.