Sentiment Lexicon Research Articles

Incremental Word Vectors for Time-Evolving Sentiment Lexicon Induction

A sentiment lexicon is a list of expressions annotated according to affect categories such as positive, negative, anger and fear. Lexicons are widely used in sentiment classification of tweets, especially when labeled messages are scarce. Sentiment lexicons are prone to obsolescence due to: 1) the arrival of new sentiment-conveying expressions such as #trumpwall and #PrayForParis and 2) temporal changes in sentiment patterns of words (e.g., a scandal associated with an entity). In this paper, we propose a methodology for automatically inducing continuously updated sentiment lexicons from Twitter streams by training incremental word sentiment classifiers from time-evolving distributional word vectors. We experiment with various sketching techniques for efficiently building incremental word context matrices and study how the lexicon adapts to drastic changes in the sentiment pattern. Change is simulated by randomly picking some words from a testing partition of words and swapping their context with the context of words exhibiting the opposite sentiment. Our experimental results show that our approach allows for successfully tracking of the sentiment of words over time even when drastic change is induced.

Stock Price Prediction Incorporating Market Style Clustering

Market style analysis is critical when designing a stock price prediction framework. Under different market styles, stocks may show quite different behaviors; thus, predictions will vary. Consequently, incorporating market styles into stock price predictions should help improve the prediction performance. In this paper, we investigate how to characterize market styles to improve stock prediction performance under varying market styles. First, stock time series data are divided into windows of different lengths. The windows are summarized and represented by technical indicators and news sentiment features. Second, hierarchical clustering is employed to cluster the windows and categorize their market styles; the window lengths and number of market styles are carefully tuned to achieve the best clustering results. Third, a distance measurement is proposed to distinguish among rotating patterns within the market styles to verify the usability of the market styles. Finally, a stock price prediction framework is constructed to predict future stock price trends based on data belonging to the same market styles. The experiments are conducted with five years of real Hong Kong Stock Exchange data that includes both stock prices and corresponding news. Two famous sentiment dictionaries (i.e., SenticNet 5 and the Loughran-McDonald financial sentiment dictionary 2018) are employed to analyze the news sentiments. Predictive models are compared both with and without incorporating market styles. The results demonstrate that the approach incorporating market styles outperforms the baseline, which does not incorporate market styles. There is a maximum 9 percent improvement in terms of accuracy and F1-score. Moreover, backtesting results show that incorporating market styles into trading signals earns trading strategies more profits on most stocks.

Sentiment Classification Algorithm Based on the Cascade of BERT Model and Adaptive Sentiment Dictionary

The mobile social network contains a large amount of information in a form of commentary. Effective analysis of the sentiment in the comments would help improve the recommendations in the mobile network. With the development of well‐performing pretrained language models, the performance of sentiment classification task based on deep learning has seen new breakthroughs in the past decade. However, deep learning models suffer from poor interpretability, making it difficult to integrate sentiment knowledge into the model. This paper proposes a sentiment classification model based on the cascade of the BERT model and the adaptive sentiment dictionary. First, the pretrained BERT model is used to fine‐tune with the training corpus, and the probability of sentiment classification in different categories is obtained through the softmax layer. Next, to allow a more effective comparison between the probabilities for the two classes, a nonlinearity is introduced in a form of positive‐negative probability ratio, using the rule method based on sentiment dictionary to deal with the probability ratio below the threshold. This method of cascading the pretrained model and the semantic rules of the sentiment dictionary allows to utilize the advantages of both models. Different sized Chnsenticorp data sets are used to train the proposed model. Experimental results show that the Dict‐BERT model is better than the BERT‐only model, especially when the training set is relatively small. The improvement is obvious with the accuracy increase of 0.8%.

GINS: A Global intensifier-based N-Gram sentiment dictionary

Words are one of the most essential elements of expressing sentiments in context although they are not the only ones. Also, syntactic relationships between words, morphology, punctuation, and linguistic phenomena are influential. Merely considering the concept of words as isolated phenomena causes a lot of mistakes in sentiment analysis systems. So far, a large amount of research has been conducted on generating sentiment dictionaries containing only sentiment words. A number of these dictionaries have addressed the role of combinations of sentiment words, negators, and intensifiers, while almost none of them considered the heterogeneous effect of the occurrence of multiple linguistic phenomena in sentiment compounds. Regarding the weaknesses of the existing sentiment dictionaries, in addressing the heterogeneous effect of the occurrence of multiple intensifiers, this research presents a sentiment dictionary based on the analysis of sentiment compounds including sentiment words, negators, and intensifiers by considering the multiple intensifiers relative to the sentiment word and assigning a location-based coefficient to the intensifier, which increases the covered sentiment phrase in the dictionary, and enhanced efficiency of proposed dictionary-based sentiment analysis methods up to 7% compared to the latest methods.

ԲԱՌԱՊԱՇԱՐԻ ՁԵՎԱՎՈՐՄԱՆ ՍԿԶԲՈՒՆՔՆԵՐԸ ՀՈՒԶԱՐՏԱՀԱՅՏՉԱԿԱՆՈՒԹՅԱՆ ՎԵՐՀԱՆՄԱՆ ԾՐԱԳՐԱԿԱԶՄՈՒՄ / THE PRINCIPLES OF SENTIMENT LEXICON GENERATION IN SENTIMENT ANALYSIS SOFTWARE

Sentiment analysis (SA) is an essential part of computational linguistics that aims at disclosing the polarity of the text. This process is mainly based on singling out sentiment lexicon since words and phrases are the main means of conveying sentiments. The main objective of this article is to present the principles of sentiment lexicon generation. The latter is mainly carried out through three approaches, namely manual approach, dictionary-based approach, and corpus-based approach. Each approach has its own peculiarities and can be used to make sentiment lexicon compilation accurate and relevant. Those approaches are intertwined and they complement each other perfectly. The topicality of the research is mainly determined by the development of information technology that promotes the dissemination of online content pertaining to several sentiments.

Bullet Subtitle Sentiment Classification Based on Affective Computing and Ensemble Learning

The bullet subtitle reflects a kind of instant feedback from the user to the current video. It is generally short but contains rich sentiment. However, the bullet subtitle has its own unique characteristics, and the effect of applying existing sentiment classification methods to the bullet subtitle sentiment classification problem is not ideal. First, since bullet subtitles usually contain a large number of buzzwords, existing sentiment lexicons are not applicable, we propose Chinese Bullet Subtitle Sentiment Lexicon on the basis of existing sentiment lexicons. Second, considering that some traditional affective computing methods only consider the text information and ignore the information of other dimensions, we construct a bullet subtitle affective computing method by combining the information of other dimensions of the bullet subtitle. Finally, aiming at the problem that existing classification algorithms ignore the importance of sentiment words in short texts, we propose a sentiment classification method based on affective computing and ensemble learning. Our experiment results show that the proposed method has higher accuracy and better practical application effect.

How Much Can Machines Learn Finance From Chinese Text Data?

Most studies on equity markets using text data focus on English-based specified sentiment dictionaries or topic modeling. However, can we predict the impact of news directly from the text data? How much can we learn from such a direct approach? We present here a new framework for learning text data based on the factor model and sparsity regularization, called FarmPredict, to let machines learn financial returns automatically. Unlike other dictionary-based or topic models that have stringent pre-screening processes, our framework allows the model to extract information more fully from the whole article. We demonstrate our study on the Chinese stock market, as Chinese text has no natural spaces between words and phrases and the Chinese market has a very large proportion of retail investors. These two specific features of our study differ significantly from the previous literature that focuses on English-text and the U.S. market. We validate our method using the literature on the Chinese stock market with several existing approaches. We show that positive sentiments scored by our FarmPredict approach generate on average 83 bps stock daily excess returns, while negative news has an adverse impact of 26 bps on the days of news announcements, where both effects can last for a few days. This asymmetric effect aligns well with the short-sale constraints in the Chinese equity market. As a result, we show that the machine-learned sentiments do provide sizeable predictive power with an annualized return of 116% with a simple investment strategy and the portfolios based on our model significantly outperform other models. This lends further support that our FarmPredict can learn the sentiments embedded in financial news. Our study also demonstrates the far-reaching potential of using machines to learn text data.

Data Augmentation for Sentiment Analysis Using Sentence Compression-Based SeqGAN With Data Screening

Sentiment analysis refers to the process of automatically identifying the emotions expressed by people. Its accuracy is highly dependent on the amount of training data. However, it takes time and cost for humans to collect a large number of data. Many research works used generative models to generate a large amount of data based on a small amount of data for sentiment analysis. However, training on long texts and inaccurate sentiment information that might be generated are two severe challenges. It is difficult to improve the sentiment analysis accuracy effectively. In this paper, we propose a novel data augmentation framework based on Sequence generative adversarial networks (SeqGAN) to improve the sentiment analysis accuracy when the dataset already has a certain amount of data and contains long texts. Penalty-based SeqGAN is used to generate high-quality and diversified text data. Long short-term memory (LSTM) networks with attention mechanisms are used to conduct sentence compression for the training data of SeqGAN. A sentiment dictionary is used to retain the sentiment words for compressed data. We also propose a data screening method to obtain more accurate data from the generated data. The results of the usability, novelty, and diversity of the generated data show that the proposed sentence compression method can help SeqGAN learn more information from the long text data. The data generated by the proposed framework improve the classification accuracy of four classifiers applied on two distinct text datasets.

Refined Global Word Embeddings Based on Sentiment Concept for Sentiment Analysis

Sentiment Analysis is an important research direction of natural language processing, and it is widely used in politics, news and other fields. Word embeddings play a significant role in sentiment analysis. The existing sentiment embeddings methods directly embed the sentiment lexicons into traditional word representation. This sentiment representation methods can only differentiate the sentiment information of different words, not the same word in different contexts, so it cannot provide accurate sentiment information for word in different contexts. This paper proposes sentiment concept to solve the problem. First, we found the optimal sentiment concept of words in Microsoft Concept Graph according to the context of words. Then we obtained the sentiment information of words under optimal sentiment concept from the multi-semantics sentiment intensity lexicon which we constructed in this paper to achieve accurate embedding of sentiment information and provide more accurate semantics and sentiment representation for words. Finally, we combined two refined word embeddings methods to achieve a more comprehensive word representation. Compared with traditional and sentiment embeddings methods on six representative datasets, the validity of the word embeddings method based on sentiment concept proposed in this paper is verified.

Determining the Efficiency of Drugs Under Special Conditions From Users’ Reviews on Healthcare Web Forums

Sentiment analysis is the extraction and categorization of sentiments that have been expressed in text data using text analysis techniques. Manifested by earlier studies, sentiment analysis of drug reviews has a large potential for providing valuable insights to assist healthcare professionals and companies for evaluating the safety of drugs after it has been marketed. Such insights help safeguard patients and increase their trust in medical companies. The existing systems either follow a lexicon-based approach or a learning-based approach for sentiment analysis in the medical domain. Learning-based techniques require annotated data while lexicon-based techniques tend to be domain-specific which restricts their wide use. This research embarks on a hybrid technique that utilizes both learning-based and lexicon-based approaches to achieve better results. General-purpose sentiment lexicons, such as AFFIN, TextBlob, and VADER, are used for annotating the reviews. Furthermore, several feature engineering techniques, such as term frequency (TF), term frequency-inverse document frequency (TF-IDF), and union of TF and TF-IDF (TF U TF-IDF) have been incorporated for the extraction of useful features. Finally, the learning models including logistic regression (LR), AdaBoost classifier (AB), random forest (RF), extra tree classifier (ETC), and multilayer perceptron (MLP) are used to classify sentiments of the reviews. The performance of the proposed hybrid approach is evaluated using accuracy, precision, recall, and F1-score. Experimental results indicate that the combination of learning-based and lexicon-based approaches provide improved results than their individual use. Moreover, TextBlob has shown promising results giving an accuracy of 96% with MLP when used with TF-IDF and with LR when used with TF U TF-IDF.

A Semantic Conceptualization Using Tagged Bag-of-Concepts for Sentiment Analysis

Sentiment could be expressed implicitly or explicitly in the text. Hence, it is the main challenge for current sentiment analysis (SA) approaches to identify hidden sentiments, other common challenges include false classification of opinion words, ignoring context information, and bad handling of a short text that arise from the bad interpretation of the text and lack of enough data required for analysis tasks. In this study, a semantic conceptualization method using tagged bag-of-concepts for SA is proposed to detect the correct sentiment towards the actual target entity that considers all affective and conceptual information conveyed in the text with a special focus on the short text. Tagged bag-of-concepts (TBoC) is a novel approach to analyze and decompose text to uncover latent sentiments while preserving all relations and vital information to boost the accuracy of SA. This study answers questions: Does the information provided via TBoC enhance sentiment classification results on different analysis levels? Is building a structure of concepts increases the accuracy of overall sentiment towards specific opinion target? Does TBoC approach enhance SA results for short text messages? The proposed solution has been applied on two datasets from the restaurant domain, sentiment analysis is performed using the TBoCs structure on multiple levels including document, aspect, aspect-category, and topic levels. TBoC method with domain-specific sentiment lexicon showed exceptional performance and outperformed other state-of-the-art NB, SVM, and NN methods, especially for aspect-level SA. The use of TBoC within the semantic conceptualization model that leverages NLP tasks, Ontology, and semantic methods proved its high capabilities for concept extraction while preserving the information about the context, interrelations, and latent feelings.

Intelligent Hybrid Feature Selection for Textual Sentiment Classification

Sentiment Analysis (SA) aims to extract useful information from online Unstructured User-Generated Contents (UUGC) and classify them into positive and negative classes. State-of-the-art techniques for SA suffer a high dimensional feature space because of noisy and irrelevant features from the UUGC. Researchers have also proposed feature extraction and selection techniques to reduce high dimensional feature space, but they fall short in extracting and selecting the most effective sentiment features for sentiment model learning. Effective feature extraction and selection are significant for the SA because they can boost the learning algorithm’s predictive performance while reducing the high-dimensional feature space. To address these concerns, we propose an Intelligent Hybrid Feature Selection for Sentiment Analysis (IHFSSA) based on ensemble learning methods. IHFSSA first identifies sentiment features in the review text utilizing Penn Treebank part-of-speech tagset and integrated Wide Coverage Sentiment Lexicons (WCSL). The sentiment features subset is then selected employing a fast and simple rank-based ensemble of multiple filters feature selection method. The selected sentiment features are further refined by applying a wrapper-based backward feature selection method. Finally, for textual sentiment classification, the well-known classification algorithms Support Vector Machine (SVM), Naive Bayes (NB), Generalized Linear Model (GLM) are trained in the ensemble model on the refined sentiment feature set. The in-depth evaluation using heterogeneous domain benchmark datasets demonstrates that IHFSSA outperforms existing SA techniques.

Efficient Sentiment-Aware Web Crawling Methods for Constructing Sentiment Dictionary

In traditional web crawling, all web pages crawled are first stored to databases. As a result, this approach can store unnecessary web pages and requires additional running time for the construction of a sentiment dictionary in a particular domain because sentiment words should be identified by scanning all web pages in the database. To address these problems, we first define the sentiment-aware web crawling problem and then propose two hash-based methods for the implementation. One is based on hash join and the other is bucket-sorted hash join. In particular, we propose a novel bucket-sorted hash join for the efficient sentiment-aware web crawling method. Our experimental results show that the proposed web crawling method using bucket-sorted hash join outperforms existing web crawling methods by significantly reducing the running time and storage space. In the proposed method, the time taken to execute the sentiment-aware task per web page is 0.016 seconds and the database space can be saved by 59% compared to the existing web crawling methods.

Leveraging Affective Hashtags for Ranking Music Recommendations

Mood and emotion play an important role when it comes to choosing musical tracks to listen to. In the field of music information retrieval and recommendation, emotion is considered contextual information that is hard to capture, albeit highly influential. In this study, we analyze the connection between users` emotional states and their musical choices. Particularly, we perform a large-scale study based on two data sets containing 560,000 and 90,000 #nowplaying tweets, respectively. We extract affective contextual information from hashtags contained in these tweets by applying an unsupervised sentiment dictionary approach. Subsequently, we utilize a state-of-the-art network embedding method to learn latent feature representations of users, tracks and hashtags. Based on both the affective information and the latent features, a set of eight ranking methods is proposed. We find that relying on a ranking approach that incorporates the latent representations of users and tracks allows for capturing a user's general musical preferences well (regardless of used hashtags or affective information). However, for capturing context-specific preferences (a more complex and personal ranking task), we find that ranking strategies that rely on affective information and that leverage hashtags as context information outperform the other ranking strategies.

Research on the emotional polarity of Academic micro-blog knowledge transfer

[Purpose / significance] This paper takes the official tweet of the Chinese Academy of Sciences as an example to explore the emotional polarity tendency of academic tweet in knowledge transfer and the relationship between these different emotional tedecies and its influential data, so as to analyze the current situation of academic public opinion, and through the analysis of the relationship between different indicators, to provide reference and suggestions for better transfer of academic knowledge. [Method / process] Based on the extended sentiment dictionary, the sentiment analysis rules are established, and the program language is used to recognize the sentiment score of comments. After getting the total score of the comments below each tweet, the relationship between the sentiment of the tweet and its corresponding influential data is identified by using the correlation analysis method. [Results / Conclusion] First of all, the comments under the tweet published by the Chinese Academy of sciences are mainly positive emotions, and the overall emotions of the tweet are also positive, but there are also some negative public opinions. Secondly, there is a moderate or above positive correlation between the overall emotion of the tweet and its influential data (number of Forwards, number of Comments, number of Likes), and the greater the number of the influential data is, the higher the emotional positivity of the original text will be. Finally, in the group of indicators of Forward--Comment, they always present a high positive correlation, while in the two groups of indicators of Forward--Like and Comment--Like, the difference in the number of samples will have a great impact on the results, and with the increase of the number of samples, the correlation is decreasing.

Analyzing Vocabulary Characteristics of Colloquial Style Corpus and Automatic Construction of Sentiment Lexicon

모바일 환경에서 의사소통은 SMS 문자로 이루어진다. SMS 문자에서 사용되는 어휘들은 일반적인 한국어 문어체 문장에서 사용되는 어휘들과 다른 부류의 어휘들이 사용될 것으로 예상할 수 있다. 예를 들어, 일반적인 문어체의 경우 문장의 시작이나 끝맺음이 올바르고 문장의 구성요소가 잘 갖추어졌지만, SMS 문자 말뭉치의 경우 구성요소를 생략 및 간략한 표현으로 대체하는 경우가 많다. 이러한 어휘 사용 특성을 분석하기 위하여, 기존에 구축된 구어체 말뭉치와 문어체 말뭉치를 사용한다. 실험에서는 구어체 말뭉치인 SMS 문자 말뭉치와 네이버 영화평 말뭉치, 그리고 문어체 말뭉치인 한국어 문어체 원시 말뭉치의 어휘 사용 특성을 비교-분석한다. 말뭉치별 어휘 비교 및 분석을 위하여 품사 태그 형용사(VA)를 기준으로 하였고, 공연강도를 측정하기 위해 변별적 공연어휘소 분석 방법론을 사용하였다. 그 결과 ‘좋-’, ‘죄송하-’, ‘즐겁-’ 등 감정표현 형용사들이 SMS 문자 말뭉치에서 선호되는 반면, 네이버 영화평 말뭉치에서는 평가 표현과 관련된 형용사들이 선호되는 것을 확인할 수 있었다. 이러한 과정에서 추출된 공연강도가 높은 형용사를 기준으로 감정어휘 사전을 자동 구축하기 위하여 단어 임베딩 기법을 사용하였으며, 총 343,603개의 감성어휘를 자동 구축하였다.

Method for product selection considering consumer’s expectations and online reviews

Purpose In recent years, the updating speed of products has been significantly accelerated, which not only provides diversified styles for consumers to select from but also makes consumers face selection problems sometimes. In addition, a large number of online reviews for products emerge on many e-commerce websites and influence consumers’ purchasing decisions. The purpose of this study is to propose a method for product selection considering consumer’s expectations and online reviews to support consumers’ purchasing decisions. Design/methodology/approach The product attributes are divided into two categories, i.e. demand attributes and word-of-mouth (WOM) attributes. For the demand attributes, for which the consumers can give specific quantified expectations, the value function of prospect theory is used to determine the consumer’s perceived values to the alternative products according to consumers’ expectations for these attributes and products’ specifications. For the WOM attributes, for which the consumers cannot give specific quantified expectations, the sentiment analysis method is used to identify the sentiment strengths for these attributes in the online reviews, and then the consumer’s perceived values to the alternative products are determined. On this basis, the product selection methods for single consumers and group consumers are given respectively. Findings Finally, taking the data of JD.com (https://www.jd.com/) as an example, the practicability and rationality of the method proposed in this paper is validated. Originality/value First, a new product selection problem considering consumer’s expectations and online reviews is extracted. Second, the product attributes are considered more comprehensively and are classified into two main categories. Third, the bounded rationality of the consumers in the decision-making process is described more reasonably. Fourth, the sentiment dictionaries for each WOM attribute are constructed and the algorithm step of identifying the sentiment strengths is designed, which can help to identify the sentiment strengths in the online reviews more accurately. Fifth, the situation that a group plans to purchase the same products and the members have inconsistent expectations for the product attributes is considered.

Automatic construction of domain-specific sentiment lexicon for unsupervised domain adaptation and sentiment classification

Sentiment analysis has long been suffering from inaccuracies using either machine learning methods that mostly benefit from text features or sentiment lexicon-based methods that are prone to domain-dependent problems. Furthermore, since labeling is a time-consuming and an expensive task, supervised machine learning methods suffer from the drawback of insufficient labeled samples. To tackle the mentioned issues, this paper proposes a novel approach with a hybrid of a neural network and a sentiment lexicon. This combination can simultaneously adapt word polarities to the target domain and leverage the polarity of whole document in order to alleviate the need for large labeled corpora in an unsupervised manner. In this respect, a sentiment lexicon is constructed from the source domain in the preprocessing phase using the labeled data. In the Next phase, having a Multilayer Perceptron (MLP), the weights of the first hidden layer are set to the corresponding polarity of each word from the retrieved sentiment lexicon and the network is trained. Finally, a Domain-Independent Lexicon (DIL) is introduced which contains words (mostly adjectives) with static positive or negative scores independent from a specific domain. After feeding the target domain to the pre-trained model, the total accuracy of the framework is enhanced by estimating the sentiment polarity of each sentence using the summation of the scores of the constitutive domain independent words. The experiments on Amazon multi-domain sentiment dataset illustrate that our approach significantly outperforms several alternative previous approaches of unsupervised domain adaptation.

Research on sentiment classification of Chinese short text

Sentiment classification can be widely used in various fields because of the value it has gained, and has received extensive attention in recent years. No matter the traditional methods based on language rules, machine statistical learning, or the current popular deep learning have achieved great success in this respect, but up to now, because of the particularity of Chinese, there are still few methods for Chinese text sentiment classification. Moreover, the traditional method is time-consuming and labor-consuming, with poor versatility. The neural network method saves resources and is generally applicable, but the application of language knowledge is not sufficient. For this reason, we propose a method combining traditional method with neural network, which combines sentiment lexicon, attention mechanism and stacked BILSTM. Word vectors are generated using word2vec, and the context semantics are captured multiple times by stacked BILSTM to get richer semantic information, for sentiment classification, sentiment words often have more influence on the result of classification, so we use sentiment lexicon which contains rich sentiment word to monitor the attention mechanism, and make the attention model pay more attention to sentiment words. Combining the advantages of traditional methods and neural networks, we use the proposed method to do the sentiment binary classification experiments on two data sets. The experimental results show that our proposed method is superior to the previous method.

Latent Semantic Scaling: A Semisupervised Text Analysis Technique for New Domains and Languages

ABSTRACT Many social scientists recognize that quantitative text analysis is a useful research methodology, but its application is still concentrated in documents written in European languages, especially English, and few sub-fields of political science, such as comparative politics and legislative studies. This seems to be due to the absence of flexible and cost-efficient methods that can be used to analyze documents in different domains and languages. Aiming to solve this problem, this paper proposes a semisupervised document scaling technique, called Latent Semantic Scaling (LSS), which can locate documents on various pre-defined dimensions. LSS achieves this by combining user-provided seed words and latent semantic analysis (word embedding). The article demonstrates its flexibility and efficiency in large-scale sentiment analysis of New York Times articles on the economy and Asahi Shimbun articles on politics. These examples show that LSS can produce results comparable to that of the Lexicoder Sentiment Dictionary (LSD) in both English and Japanese with only small sets of sentiment seed words. A new heuristic method that assists LSS users to choose a near-optimal number of singular values to obtain word vectors that best capture differences between documents on target dimensions is also presented.