Emotion Lexicon Research Articles

Mining Emotions from the Finnish War Letter Collection, 1939– 1944

Our paper analyses emotional language used by Finnish soldiers and civilians in their private communication during World War II. The dataset consists of 7,000 handwritten letters converted into a machine-readable corpus with rich metadata. The dataset offers a unique opportunity to make a statistical analysis of people’s emotional responses to the war. We engage in key questions of the cultural history of war, such as the connection between soldiers’ emotional language and violence: did soldiers’ emotional language become more laconic in the course of the war? While computational approaches to mining emotions have been common in fields like computer science and linguistics, they have not gained wider popularity in historical research. Pioneering attempts have been based on individual emotion words carefully chosen by an historian, or on readily available, more generic emotion lexicons. Compared to machine-learning solutions, lexicon-based approaches require less computational effort and are more transparent to interpret. Our methodology combines the ready-made word list FEIL with contextual knowledge of historians. FEIL gives around 7,000 Finnish words an emotion category and intensity ratings. First, the emotion lexicon was filtered based on high intensity. Then the domain expert manually removed words not particularly emotionally intensive in the context of war letters. The expert also annotated the list of the most frequent words in the war letter collection and handpicked emotionally intensive words not included in FEIL. Our final list covered 298 emotion words. We quantified changes in their use over time. In contrast to earlier research, our analysis indicates that soldiers’ and civilians’ emotionality did not significantly differ during World War II. Soldiers’ use of emotion words saw a decline in the last stages of the war, but overall their letters were almost as emotional as the civilians’ letters. We did indeed identify some changes in the individual emotion words used by the soldiers in their letters: patriotic words in particular decreased in the course of the war. In addition to empirical findings, our paper sheds light on the problem of universal emotion lexicons in historical research: linguistic, cultural and temporal differences between present-day lexicons and historical datasets can lead to biased interpretations. Thus, our paper contributes not only to the history of emotions but also to emotion mining, which is historically sensitive.

Stock Market Prediction Based on Financial News Text Mining and Investor Sentiment Recognition

The stock market is usually regarded as the bellwether of the economy, which can reflect the economic operation of a country or region. As a significant part of the financial market, the equity market plays a critical role in the financial sector. Whether in academia or investment field, stock market forecasts always excite great interest. Financial news is an important source of information in the financial market, which reflects the mood swings of investors and often goes hand in hand with the market trend. However, due to the unstructured and professional characteristics of financial news, there are challenges in accurately quantifying their emotional tendencies. This research is based on Hidden Markov Model (HMM) to segment financial news text. The recognition and classification of news emotion is carried out by bidirectional long short-term memory (BI-LSTM) algorithm, and long short-term memory(LSTM) model is trained with text emotion index and stock market transaction data to realize the prediction of stock market. The results show that BI-LSTM algorithm performs better than the emotional dictionary algorithm in emotional recognition. And the emotional index of financial news text can enhance the accuracy of stock market prediction to a certain extent. Compared with using stock market technical index and news text vector only, the prediction accuracy can be improved by about 2%.

A Deep Learning-Based Text Emotional Analysis Framework for Yellow River Basin Tourism Culture

As an important carrier of cultural communication, tourism can play a positive role in promoting regional ecology and cultural heritage. Therefore, this paper takes tourist attractions in the Yellow River basin as the research object and constructs mining and comment sentiment analysis for tourism text information in the Yellow River basin that appears on social media platforms. Based on the theory of the social center network, the tourism culture network of the Yellow River basin based on tourists’ emotion analysis is constructed. In addition, based on the linear fusion algorithm of semantic orientation pointwise mutual information and word2vec, this paper constructs an emotion dictionary in the field of tourism review and proposes a set of comprehensive emotion calculation rules based on Chinese text expression structure. The experimental results of 32 scenic spots in the Yellow River basin show that the proposed algorithm can achieve better sentiment classification of tourism texts, broaden the scope of application of the domain sentiment dictionary construction method, and improve efficiency.

The Public Sentiment Analysis of Double Reduction Policy on Weibo Platform.

Weibo platform is an indispensable transmission channel in education policy release and dissemination. The events and sentiments contained in education policies microblogs include the public sentiment and support the general management and guidance scientifically and efficiently. This study constructs a dataset based on the “Double Reduction Policy” relevant microblogs and comments. The policy events are extracted by Latent Dirichlet Allocation (LDA) model and Language Technology Platform (LTP). Based on the emotion dictionary, an attention-based BiLSTM model is constructed to classify the public sentiments. The experimental results reveal four themes: “industry impact,” “institutional supervision,” “public feedback,” and “policy implementation.” The distribution conforms to the development trend of online public sentiments.

Strategic sentiments and emotions in post-Second World War party manifestos in Finland

We contribute to the growing number of studies on emotions and politics by investigating how political parties strategically use sentiments and emotions in party manifestos. We use computational methods in examining changes of sentiments and emotions in Finnish party manifestos from 1945 to 2019. We use sentiment and emotion lexicons first translated from English into Finnish and then modified for the purposes of our study. We analyze how the use of emotions and sentiments differs between government and opposition parties depending on their left/right ideology and the specific type of party manifesto. In addition to traditional sentiment and emotion analysis, we use emotion intensity analysis. Our results indicate that in Finland, government and opposition parties do not differ substantially from each other in their use of emotional language. From a historical perspective, the individual emotions used in party manifestos have persisted, but changes have taken place in the intensity of using emotion words. We also find that in comparison with other parties, populist parties both appeal to different emotions and appeal to the same emotions with different intensities.

Analysis of Ensemble Classification of Twitter Sentiments Using New Dependency Tree Based Approach

The vast amount of data available on social media and microblogs can be a valuable resource for mining opinions or for analyzing the overall mood of the public. This helps in identifying potential customers, exploring market trends and predicting events. Analyzing twitter data is comparatively difficult due to the large amount of irregularities present in tweets. Many approaches that use sentiment dictionaries and machine learning have been proposed until now. In this paper, we present a new feature that is extracted using dependency parsing and an emotion lexicon. This feature, along with n-grams, syntactic n-grams and lexicon-based features, is used to classify the tweets. We also use custom dictionaries to identify slang words, SMS short forms, emoticons and word contractions. The performance of various classification algorithms and ensemble techniques is compared. Our results show that the new feature along with the ensemble framework improves sentiment classification.

Multilabel Emotion Tagging for Domain-Specific Texts

In this article, we propose a novel approach to bootstrap a general seed emotion lexicon with words found in a domain-specific corpus. The approach divulges the contextual similarity between two words in the corpus via lexical-, dictionary-, and topic-based features, thus revealing the emotion labels of domain-specific words. As unfolding the recursive structure of language is an irreducible component of how humans understand a sentence, in this study, a propagation mechanism is designed that takes advantage of a shallow parser to derive the emotions associated with the words and their parent phrases. This mechanism pushes beyond the limits of most word co-occurrence approaches and facilitates the multilabel emotion tagging of a sentence in a manner reflecting human cognition. Evaluations on two benchmark corpora support the validity of the propagation mechanism. Further evaluation of a financial corpus indicates that our system outperforms the traditional bag-of-words approach. Our approach provides better modeling of compositional emotions by considering the emotion-bearing words, shifters, intensifiers, and overall sentence structure.

Weighted Joint Sentiment-Topic Model for Sentiment Analysis Compared to ALGA: Adaptive Lexicon Learning Using Genetic Algorithm.

Latent Dirichlet Allocation (LDA) is an approach to unsupervised learning that aims to investigate the semantics among words in a document as well as the influence of a subject on a word. As an LDA-based model, Joint Sentiment-Topic (JST) examines the impact of topics and emotions on words. The emotion parameter is insufficient, and additional parameters may play valuable roles in achieving better performance. In this study, two new topic models, Weighted Joint Sentiment-Topic (WJST) and Weighted Joint Sentiment-Topic 1 (WJST1), have been presented to extend and improve JST through two new parameters that can generate a sentiment dictionary. In the proposed methods, each word in a document affects its neighbors, and different words in the document may be affected simultaneously by several neighbor words. Therefore, proposed models consider the effect of words on each other, which, from our view, is an important factor and can increase the performance of baseline methods. Regarding evaluation results, the new parameters have an immense effect on model accuracy. While not requiring labeled data, the proposed methods are more accurate than discriminative models such as SVM and logistic regression in accordance with evaluation results. The proposed methods are simple with a low number of parameters. While providing a broad perception of connections between different words in documents of a single collection (single-domain) or multiple collections (multidomain), the proposed methods have prepared solutions for two different situations (single-domain and multidomain). WJST is suitable for multidomain datasets, and WJST1 is a version of WJST which is suitable for single-domain datasets. While being able to detect emotion at the level of the document, the proposed models improve the evaluation outcomes of the baseline approaches. Thirteen datasets with different sizes have been used in implementations. In this study, perplexity, opinion mining at the level of the document, and topic_coherency are employed for assessment. Also, a statistical test called Friedman test is used to check whether the results of the proposed models are statistically different from the results of other algorithms. As can be seen from results, the accuracy of proposed methods is above 80% for most of the datasets. WJST1 achieves the highest accuracy on Movie dataset with 97 percent, and WJST achieves the highest accuracy on Electronic dataset with 86 percent. The proposed models obtain better results compared to Adaptive Lexicon learning using Genetic Algorithm (ALGA), which employs an evolutionary approach to make an emotion dictionary. Results show that the proposed methods perform better with different topic number settings, especially for WJST1 with 97% accuracy at |Z| = 5 on the Movie dataset.

Affective Knowledge-enhanced Emotion Detection in Arabic Language: A Comparative Study

Online opinions/reviews contain a lot of sentiments and emotions that can be very useful, especially, for Internet suppliers which can know whether their services/products are meeting their customers’ expectations or not. To detect these sentiments and emotions, most applications resort to lexicon-based approaches. The major issue here is that most well-known emotion lexicons have been developed for English language; nevertheless, in other languages such as Arabic, there are fewer available tools, and many times, the quality of them is poor. The goal of this study is to compare the performance of two different types of algorithms, shallow machine learning-based and deep learning-based, when dealing with emotion detection in Arabic language. To improve the performance of the algorithms, two lexicons, which were originally developed in other languages and translated into Arabic language, have been used to add emotional features to different information models used to represent opinions. All approaches have been tested using the dataset SemEval 2018 Task 1: Affect in Tweets and the dataset LAMA+DINA. The semantic approaches outperform the classical algorithms, that is, the information provided by the lexicons clearly improves the results of the algorithms. Particularly, the BiLSTM algorithm outperforms the rest of the algorithms using word2vec. On the contrary to other languages, the best results were obtained using the NRC lexicon.

Explainable Deep Attention Active Learning for Sentimental Analytics of Mental Disorder

With the increasing use of online mediums, Internet-delivered psychological treatments (IDPs) are becoming an essential tool for improving mental disorders. Online-based health therapies can help a large segment of the population with little resource investment. The task is greatly complicated by the overlapping emotions for specific mental health. Early adoption of a deep learning system presented severe difficulties, including ethical and legal considerations that contributed to a lack of trust. Modern models required highly interpretable, intuitive explanations that humans could understand. To achieve this, we present a deep attention model based on fuzzy classification that uses the linguistic features of patient texts to build emotional lexicons. In medical applications, a diversified dataset generates work. Active learning techniques are used to extend fuzzy rules and the learned dataset gradually. From this, the model can gain a reduction in labeling efforts in mental health applications. In this way, difficulties such as the amount of vocabulary per class, method of generation, the source of data, and the baseline for human performance level can be solved. Moreover, this work illustrates fuzzy explainability by using weighted terms. The proposed method incorporates a subset of unstructured data into the set for training and uses a similarity-based approach. The approach then updates the model training using the new training points in the subsequent cycle of the active learning mechanism. The cycle is repeated until the optimal solution is found. At this point, all unlabeled text is converted into the set for training. The experimental results show that the emotion-based enhancement improves test accuracy and helps develop quality criteria. In the blind test, the bidirectional LSTM architecture with an attention mechanism and fuzzy classification achieved an F1 score of 0.89.

Impact du 3e confinement lié à la Covid19 sur les émotions des Français : exploration textuelle de 481 601 flux Twitter

ObjectiveThis work proposes the investigation of the immediate emotional repercussions of the announcements of the third containment in France. In order to respond to this objective, we are interested in the emotional reactions as they were communicated via the social network Twitter between March 11, 2021 and April 08, 2021. MethodAmong the tweets, 481,601 were retrieved via the rtweet library of R software. An automated emotional lexicon analysis was conducted. The data were processed according to two approaches: (i) one cross-sectionally and (ii) the other one considering longitudinally the dynamic aspect of emotions by studying the content of tweets, day after day. ResultsThe impact of the containment announcements is not trivial. If positive emotions can be observed, the impact remains mostly negative and global. Over the whole period considered, the dominant feeling was one of discomfort. It is illustrated by an experience of loneliness, feelings of tension, suffering and/or frustration. Depression and suffering are also predominant in all the tweets. The temporal analyses show that the nature and diversity of emotions vary according to the date on which the tweets were published. The reconfinement announcements are thus mainly associated with hatred but appear to be only transient. DiscussionThe results are discussed with respect to the adaptive functions of emotions. ConclusionThe analysis of the messages posted on social networks brings us information for a real-time understanding of the emotional reactions related to reconfinement announcements. In addition to examining the type of emotions mobilized, a dynamic reading of them will help to bring out the personal meaning that the confinement may have had.

How is People's Awareness of "Biodiversity" Measured? Using Sentiment Analysis and LDA Topic Modeling in the Twitter Discourse Space from 2010 to 2020.

The importance of biodiversity conservation is gradually being recognized worldwide, and 2020 was the final year of the Aichi Biodiversity Targets formulated at the 10th Conference of the Parties to the Convention on Biological Diversity (COP10) in 2010. Unfortunately, the majority of the targets were assessed as unachievable. While it is essential to measure public awareness of biodiversity when setting the post-2020 targets, it is also a difficult task to propose a method to do so. This study provides a diachronic exploration of the discourse on “biodiversity” from 2010 to 2020, using Twitter posts, combined with sentiment analysis and topic modeling, commonly used in data science. Through the aggregation and comparison of n-grams, the visualization of eight types of emotional tendencies using the NRC emotion lexicon and supplemental comparison with the machine learning model, the construction of topic models using Latent Dirichlet allocation (LDA), and the qualitative analysis of tweet texts based on these models, the analysis and classification of these unstructured tweets have been performed effectively. The results revealed the evolution of words used with “biodiversity” on Twitter over the past decade, the emotional tendencies behind the contexts in which “biodiversity” has been used, and the approximate content of tweet texts that have constituted topics with distinctive characteristics. While searching for people’s awareness through SNS analysis still has many limitations, it is undeniable that essential suggestions can be obtained. To further refine the research method, it will be crucial to improve analysts’ skills, accumulate research examples, and advance data science.Supplementary InformationThe online version contains supplementary material available at 10.1007/s42979-022-01276-w.

Construction of Mental Health Education Model for College Students Based on Fine-Grained Parallel Computing Programming

Psychological education is beneficial in maintaining the psychological health of college students, resolving psychological issues, and creating a harmonious campus and society. Mental health education for college students is a development-oriented basic education activity that contributes significantly to educational quality. To improve the accuracy of college students' mental health assessments, a model for mental health education for college students based on fine-grained parallel computing programming is proposed. This study uses a deep learning algorithm to subdivide the classification of an emotion dictionary, which can be classified by adding negative word lists, polarity conversion dictionaries, and online dictionaries, among other things, based on the theory of ecological instantaneous evaluation. It can be used for both multiclass and detailed emotion analysis. The model is more accurate in assessing the mental health of college students, according to the results of the study. The current emotional state of users can be identified, as well as signs of psychological risk, using emotional analysis of Weibo data, which will become a valuable resource for users seeking clinical psychological consultation in the future.

Highlighting smart city mirages in public perceptions: A Twitter sentiment analysis of four African smart city projects

‘Smart cities’ strategies are often promoted as a panacea to existing urban planning issues such as local economic development, public safety, traffic congestion, and infrastructure maintenance. While these initiatives continue to be increasingly sought out and publicly promoted by governments across the globe, little is known about the influence of smart city branding campaigns on public perceptions of such projects. To explore this dynamic, we conducted a sentiment analysis of tweets regarding four smart city projects in Africa using the NRC Emotion Lexicon. The analysis revealed strong and consistent positive associations with words annotated with the emotions of Anticipation, Trust, and Joy. Based on these cross-case trends, we speculate that given their underwhelming outputs, the larger smart city agenda and its accompanying strategic positive framing have contrived a potentially pervasive and illusionary sentiment of “smart city mirages” among the public consciousness.

Text mining algorithm of value orientation of civil and commercial law,regulations and reimbursement based on emotional dictionary

Aiming at the low accuracy of the traditional value-oriented text mining algorithm of civil and commercial laws, regulation also reimbursement a value-oriented text mining algorithm of civil and commercial law based on an emotional dictionary is designed. Through the civil and commercial law value orientation text word segmentation, eliminate the redundant values in the text. Select the text features of the value orientation of civil and commercial law, regulation also reimbursement and use the one hot method to represent the text vector. Calculate pad and emotional intensity, construct civil and commercial law value orientation text, and mine emotional dictionary. The confusion matrix is established to realize the text mining of the value orientation of civil and commercial law. The experimental results show that the accuracy of the designed algorithm for value oriented text mining of civil and commercial law is significantly higher than that of the traditional algorithm, which can solve the problem of low accuracy of the traditional value oriented text mining algorithm of civil and commercial law.

Creating and Comparing Dictionary, Word Embedding, and Transformer-Based Models to Measure Discrete Emotions in German Political Text

AbstractPrevious research on emotional language relied heavily on off-the-shelf sentiment dictionaries that focus on negative and positive tone. These dictionaries are often tailored to nonpolitical domains and use bag-of-words approaches which come with a series of disadvantages. This paper creates, validates, and compares the performance of (1) a novel emotional dictionary specifically for political text, (2) locally trained word embedding models combined with simple neural network classifiers, and (3) transformer-based models which overcome limitations of the dictionary approach. All tools can measure emotional appeals associated with eight discrete emotions. The different approaches are validated on different sets of crowd-coded sentences. Encouragingly, the results highlight the strengths of novel transformer-based models, which come with easily available pretrained language models. Furthermore, all customized approaches outperform widely used off-the-shelf dictionaries in measuring emotional language in German political discourse.

Emotional Lexicon in Down Syndrome.

We studied comprehension of emotion versus concrete/abstract words in Down syndrome (DS). Study 1 compared 26 participants with DS and 26 typically developing (TD) children matched on verbal ability. Results showed no difference between groups. Study 2 assessed whether chronological age (CA) and (non)verbal abilities predicted developmental trajectories of comprehension in 36 children with DS and 143 TD children. For the latter, these variables predicted comprehension of all three word types. For the former, receptive vocabulary predicted comprehension of all word types, but CA and nonverbal reasoning only predicted comprehension of concrete words. This suggests that people with DS have no specific emotional lexicon deficit. Supporting their general lexical development would help them access abstract and emotional meanings.

Brand Preference Prediction Method of Cross-Border E-Commerce Consumers Based on Potential Tag Mining

There are many brands in cross-border e-commerce platforms. Obtaining consumers’ preference for brands will help promote the development of cross-border e-commerce industry. A brand preference prediction method of cross-border e-commerce consumers based on potential tag mining is proposed. Preprocess the cross-border e-commerce brand comment information obtained, build a HowNet emotion dictionary, and calculate consumers’ emotional tendency towards the brand on this basis. The projection pursuit regression model is optimized by differential evolution algorithm to reduce the dimension of the obtained consumer brand emotion information. Mining the potential labels of the information after dimensionality reduction, combined with Bayesian personalized sorting method and paired interaction tensor decomposition method, this paper constructs a brand preference’s prediction model to predict the brand preference of cross-border e-commerce consumers. The experimental results show that the proposed method has high accuracy of brand tendency calculation results, small average absolute error of prediction results, and high model accuracy.

Readers’ affect: predicting and understanding readers’ emotions with deep learning

Emotions are highly useful to model human behavior being at the core of what makes us human. Today, people abundantly express and share emotions through social media. Technological advancements in such platforms enable sharing opinions or expressing any specific emotions towards what others have shared, mainly in the form of textual data. This entails an interesting arena for analysis; as to whether there is a disconnect between the writer’s intended emotion and the reader’s perception of textual content. In this paper, we present experiments for Readers’ Emotion Detection through multi-target regression settings by exploring a Bi-LSTM-based Attention model, where our major intention is to analyze the interpretability and effectiveness of the deep learning model for the task. To conduct experiments, we procure two extensive datasets REN-10k and RENh-4k, apart from using a popular benchmark dataset from SemEval-2007. We perform a two-phase experimental evaluation, first being various coarse-grained and fine-grained evaluations of our model performance in comparison with several baselines belonging to different categories of emotion detection, viz., deep learning, lexicon based, and classical machine learning. Secondly, we evaluate model behavior towards readers’ emotion detection assessing attention maps generated by the model through devising a novel set of qualitative and quantitative metrics. The first phase of experiments shows that our Bi-LSTM + Attention model significantly outperforms all baselines. The second analysis reveals that emotions may be correlated to specific words as well as named entities.

DRIIS: Research on Automatic Recognition of Artistic Conception of Classical Poems Based on Deep Learning

Poetry is the jewel in the crown of our country’s classical culture and has been praised and studied by countless people for thousands of years. In recent years, with the rapid development of computer technology and the leap-forward improvement of hardware computing power, natural language processing (NLP) technology has achieved remarkable results in practice. We applied NLP to the text analysis of classical poetry, proposed a set of methods to automatically recognize the artistic conception in classical poetry, and established the classical poetry artistic conception dataset for experimentation through the crawler method. In the experiment, we studied the application of different machine learning algorithms in text classification, combined such algorithms with different document vectorization methods, compared their performance on the topic classification problem of poetry, and concluded that there are some better accuracy rates under the classical machine learning framework. Comparing the effects of word-based vectors and word-based vectors, we concluded that the ancient poetry word vectors constructed based on characters have a higher accuracy rate. We also further introduced deep learning methods into the research, analyzed the pros and cons of various neural networks, and studied the neural network architectures that have good results in the practice of NLP, such as TextCNN and BiLSTM models. We also introduced mature NLP pre-training models such as BERT to classify the artistic conception of classical poetry. In addition, we also constructed an emotional dictionary matching method based on word vectors for sentiment analysis. The experimental results have shown that the method proposed in this paper has a good effect of automatic recognition of classical poetry mood, which can be used to recommend similar poems and select poems with emotion as the theme through the poetry mood.