Search

Despite numerous studies that adopt text analysis in tourism and hospitality published in the past decade, a comprehensive review of this topic is still lacking. This study endeavors to bridge this gap by conducting a systematic review of text analysis research published in the top five journals in tourism and hospitality between 2013 and 2023. From a thematic analysis perspective, this paper provides an elaborate description of how text data can be utilized for prediction and understanding purposes. Specifically, the prediction scenarios encompass hotels, destinations, and restaurants, while the understanding aspect focuses on tourists, firms, destinations, and academic researches within this domain. From a methodological perspective, this paper offers an in-depth exploration of data sources, tools or methodologies, processing procedures, and commonly utilized metrics in text analysis. The paper concludes by proposing a summary framework for text analysis research while highlighting future research trends, opportunities, and challenges in this domain. [ABSTRACT FROM AUTHOR]

The digital age has resulted in a massive increase in the amount of available textual data, including articles, comments, texts, and updates on social networks. The value of analyzing such a large volume of data extends to many other industries and applications, as it provides important insights into the perspectives of customers, strategic decision-making, and market demands. Detecting emotions in texts faces challenges due to linguistic patterns and cultural nuances. This study proposes a system capable of accurately identifying emotions expressed in text using a variety of machine learning models, including logistic regression, extra randomized tree, voting, SGD, and LinearSVC. It also employs different feature extraction techniques, such as TF-IDF, Bag-of-Words, and N-grams, comparing their performance in these models. An evaluation was carried out using two English emotion datasets, namely ISEAR and AIT-2018, using F1 score, accuracy, recall, and precision. The findings demonstrate the ability and effectiveness of the system to detect emotions conveyed within texts. The LinearSVC model with N-grams achieved the highest accuracy of 88.63% on the ISEAR dataset, while the extra randomized tree classifier with N-grams achieved 89.14% accuracy on the AIT-2018 dataset. Furthermore, the SGD model with TF-IDF achieved 88.18% and 84.54% accuracy on the ISEAR and the AIT-2018 datasets, respectively. [ABSTRACT FROM AUTHOR]

Copyright of Contemporary Accounting Research is the property of Canadian Academic Accounting Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Traditional mid‐term electricity forecasting models rely on calendar and meteorological information such as temperature and wind speed to achieve high performance. However depending on such variables has drawbacks, as they may not be informative enough during extreme weather. While ubiquitous, textual sources of information are hardly included in prediction algorithms for time series, despite the relevant information they may contain. In this work, we propose to leverage openly accessible weather reports for electricity demand and meteorological time series prediction problems. Our experiments on French and British load data show that the considered textual sources allow to improve overall accuracy of the reference model, particularly during extreme weather events such as storms or abnormal temperatures. Additionally, we apply our approach to the problem of imputation of missing values in meteorological time series, and we show that our text‐based approach beats standard methods. Furthermore, the influence of words on the time series' predictions can be interpreted for the considered encoding schemes of the text, leading to a greater confidence in our results. [ABSTRACT FROM AUTHOR]

Sentiment analysis is a crucial discipline that focuses on the interpretation of feelings and points of view in textual data. Our study aims to assess the impact of different feature extraction methods on the accuracy of opinion research models. Techniques such as bag-of-words (BoW), term frequencyinverse document frequency (TF-IDF), Word2Vec, global vectors (GloVe) and bidirectional encoder representations from transformers (BERT) were used with three machine learning algorithms and three deep learning networks as classifiers. The IMDB movie review dataset was used for evaluation. The results showed that combining BERT with LSTM, CNN and RNN improved performance, achieving an accuracy rate of 94%, precision of 94.14%, recall of 93.27% and an F1 score of 89.33%. These results highlight the significant contribution of ERTB to model performance, outperforming other feature extraction techniques in text classification. The study concludes that the fusion of BERT and LSTM significantly improves model accuracy for opinion retrieval, recommending BERT as the main feature extraction method for optimizing performance in NLP tasks. [ABSTRACT FROM AUTHOR]

Problem definition: Commodity prices have exhibited significant volatility in recent times, which poses an exogenous risk factor for commodity-processing and commodity-trading firms. Accurate commodity price forecasts can help firms leverage data-driven procurement policies that incorporate the underlying price volatility for financial and operational hedging decisions. However, historical prices alone are insufficient to obtain reasonable forecasts because of the extreme volatility. Methodology/results: Building on the hypothesis that commodity prices are driven by real-world events, we propose a method that automatically extracts events from news articles and combines them with price data using a neural network-based predictive model to forecast prices. In addition to achieving a high prediction accuracy that outperforms several benchmarks (by up to 13%), our proposed model is also interpretable, which allows us to identify meaningful events driving the price fluctuations. We found that the events frequently associated with major fluctuations in the price include "natural," "hike," "policy," and "elections," all of which are known drivers of price change. We used a corpus containing about 1.6 million news articles of a major Indian newspaper spanning 15 years and daily prices of four crops (onion, potato, rice, and wheat) in India to perform this study. Our proposed approach is flexible and can be used to predict other time series data, such as disease incidence levels or macroeconomic indicators, that are also influenced by real-world events. Managerial implications: Firms can leverage price forecasts from our system to design inventory and procurement policies in the face of uncertain commodity prices. Commodity merchants can also use the forecasts to design optimal storage policies for physical trading of commodities when prices are volatile. Our findings can also significantly impact policymakers, who can leverage the information of impending price changes and associated events to mitigate the negative effects of price shocks. History: This paper has been accepted in the Manufacturing & Service Operations Management Frontiers in Operations Initiative. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0641. [ABSTRACT FROM AUTHOR]

Machine learning models built on behavioral and textual data can result in highly accurate prediction models, but are often very difficult to interpret. Linear models require investigating thousands of coefficients, while the opaqueness of nonlinear models makes things worse. Rule-extraction techniques have been proposed to combine the desired predictive accuracy of complex "black-box" models with global explainability. However, rule-extraction in the context of high-dimensional, sparse data, where many features are relevant to the predictions, can be challenging, as replacing the black-box model by many rules leaves the user again with an incomprehensible explanation. To address this problem, we develop and test a rule-extraction methodology based on higher-level, less-sparse "metafeatures". We empirically validate the quality of the explanation rules in terms of fidelity, stability, and accuracy over a collection of data sets, and benchmark their performance against rules extracted using the fine-grained behavioral and textual features. A key finding of our analysis is that metafeatures-based explanations are better at mimicking the behavior of the black-box prediction model, as measured by the fidelity of explanations. [ABSTRACT FROM AUTHOR]

Kadir Uludag Shanghai Mental Health Center, School of Medicine, Shanghai Jiao Tong University, Shanghai, People’s Republic of ChinaCorrespondence: Kadir Uludag, Shanghai Jiao Tong University, School of Medicine, 16 Wanping Nan Road, 600, Shanghai, 20030, People’s Republic of China, Tel +86-18401653162, Email kadiruludag@mails.ucas.ac.cnBackground: Textual data analysis has become a popular method for examining complex human behavior in various fields, including psychology, psychiatry, sociology, computer science, data mining, forensic sciences, and communication studies. However, identifying the most relevant textual parameters for analyzing complex behavior is still a challenge.Goal of Study: This paper aims to explore potential textual parameters that could be useful in analyzing behavior through complex textual data. Furthermore, we have examined the randomly generated text based on different textual parameters.Methods: To achieve this goal, we conducted a comprehensive review of the literature on textual data analysis and identified several potential topics that could be relevant, such as sentiment analysis, discourse analysis, lexical analysis, and syntactic analysis. We discuss the theoretical background and practical implications of each parameter and provide examples of how they have been used in previous research. Furthermore, we highlight the importance of considering the context in which these parameters are applied and the need for interdisciplinary collaboration to gain a deeper understanding of complex behavior through textual data analysis. Furthermore, we have provided Python code in the Supplementary Materials to facilitate a comprehensive analysis of such behaviors. In addition, to generate the text for analysis, we utilized ChatGPT 3.5 Turbo by requesting it to generate a random text of 1000 words divided into five paragraphs. Afterwards, we applied the provided Python code to analyze the randomly generated text.Conclusion: Overall, this paper provides a foundation for researchers to identify relevant textual parameters to analyze complex human behavior in their respective fields such as linguistics, sociology, psychiatry, and psychology.Keywords: chatbot, behavioral analysis, textual parameters, textual data, psychology, text analysis, linguistics, textual patterns

In modern China, industrialization has formed a critical foundation for the transition to modernization. However, the spatiotemporal evolution patterns and driving mechanisms of urban industrial development in Nanjing from 1840 to 1949 remain unclear. Based on textual historical sources, this study examined the spatiotemporal patterns of urban industrial development in Nanjing from 1840 to 1949 by using spatial analysis methods, GeoDetector, regression models and industrial structure indices. The results reveal the following: (1) The overall spatial distribution pattern of the industry in modern Nanjing exhibited a "one main, one secondary" dual-center "ladle-shaped" arrangement. Over time, industry has expanded from the urban center toward the east and north. (2) The modernization level of different industries was uneven, exhibiting a "center-periphery" spatial pattern. (3) At the micro level, transportation and population density were the primary influencing factors for industrial location, whereas at the macro level, government intervention mainly affected the industrialization pattern. (4) The industrial development pattern in modern Nanjing, in alignment with the "pole-axis" spatial system, serves as a microcosm of China's urban modernization transition. This study represents the application of GIS methods in the humanities and provides valuable insights for urban planning and development. [ABSTRACT FROM AUTHOR]

User feedback in text classification serves multiple purposes, including refining models, enhancing datasets, adapting to user preferences, identifying emerging topics, and evaluating system performance, all of which contribute to the creation of more effective and user-centric classification systems. Many text classification techniques, including data mining, machine learning, and deep learning approaches, have been employed in previous literature, each making significant contributions to the field. This paper aims to contribute by guiding researchers seeking commonly used classification techniques and evaluation metrics in text processing. Additionally, it identifies the classification technique that generates higher accuracy and works as a basis for researchers to synthesize studies within their respective fields. Preferred reporting items for systematic reviews and meta-analysis (PRISMA) methodology is adapted to systematically review 28 current literatures on text classification on user feedback. The results obtained are guided by four research questions; paper distribution year, dataset source and size; evaluation metric and model accuracy. The review has shown that support vector machines (SVM) are frequently employed and consistently achieve high levels of accuracy as high as 97.17% with various datasets used. The future direction of this work could explore models that integrate sentiment analysis and natural language understanding to more accurately capture nuanced user opinions and preferences. [ABSTRACT FROM AUTHOR]

Sentiment analysis, which aims to extract information from textual data indicating people's ideas or attitudes about a particular problem, has developed into one of the most exciting study issues in natural language processing (NLP) with the development of social media. Twitter is a social network with an extensive audience that expresses their thoughts and opinions clearly and readily. Due to the prevalence of slang phrases and incorrect spellings in short phrase styles, Twitter data analysis is more challenging than data analysis from other social networks. Automated feature selection still has several limitations, such as higher computing costs that rise with the number of characteristics. Deep learning, which is self-learned and more accurate at processing vast amounts of data, is utilized to overcome these challenges. This paper introduces a conditional generative adversarial network (GAN) for Twitter sentiment analysis, whereas a convolutional neural network (CNN) has been used to extract traits from Twitter data. Compared to existing works, the proposed work has outperformed in accuracy, recall, precision, and F1 score. The suggested method is the most accurate, with a classification accuracy of 93.33 %.

In Indonesia, e-commerce platforms are rapidly growing as the number of visitors continues to increase. There is a product that is most often used, called Fast-Moving Consumer Goods. The customer's side wants affordable prices, so it is challenging for the product procurement team to prepare their own estimated prices and compare them quickly. In this study, a decision support system was designed by automating the product grouping process. The experiment used FMCGs textual data from various e-commerce sites and cluster data preparation with the help of Latent Semantic Analysis to reduce the TF-IDF weighting data dimension. The results showed that the effect of data preparation yielded a result of 0.092 outperforming data without stoplist and word-based stoplists available by the stopwords library, even though the execution time was 0.01 seconds longer than others. [ABSTRACT FROM AUTHOR]

Authorship classification analyzes an author's prior work to identify their writing style, a unique trait of each language and individual author. This research aims to conduct a thorough comparative analysis of various methods for classifying authorship. The study leverages two corpora: AAALitCorpus of Albanian literary texts and CCAT10 of English columns. We evaluate modelgenerated features across different configurations. The richness of the features and the breadth of the analysis provide a significant understanding of the problem, setting a new standard for comprehensive linguistic investigations across multiple languages. The study indicates that machine learning algorithms accurately discern authorial writing styles, highlighting the complexities of classifying authorship in a cross-linguistic context. [ABSTRACT FROM AUTHOR]

This study delves into Consumer Social Responsibility (CnSR) within the fashion industry, with the goal of understanding consumers' sustainable and responsible behavior across three major consumption stages: acquisition, utilization, and disposal. While "corporate" social responsibility (CSR) has been extensively studied in the literature, CnSR that sheds light on "individual consumers" has received less attention and is understudied. Using topic modeling, an unsupervised machine learning (ML) technique that uses natural language processing (NLP) in Python, this study analyzed textual data consisting of open-ended responses from 703 U.S. consumers. The analysis unveiled key aspects of CnSR in each of the consumption processes. The acquisition stage highlighted various ethical and sustainable considerations in purchasing and decision making. During the utilization phase, topics concerning sustainable and responsible product usage, environmentally conscious practices, and emotional sentiments emerged. The disposal stage identified a range of environmentally and socially responsible disposal practices. This study provides a solid and rich definition of CnSR from the perspective of individual consumers, paving the avenue for future research on sustainable consumption behaviors and inspiring the fashion industry to create goods and services that are in line with CnSR. [ABSTRACT FROM AUTHOR]

Social media has become an essential source of news for everyday users. However, the rise of fake news on social media has made it more difficult for users to trust the information on these platforms. Most research studies focus on fake news detection in the English language, and only a limited number of studies deal with fake news in resource-poor languages such as Urdu. This article proposes a globally weighted term selection approach named normalized effect size (NES) to select highly discriminative features for Urdu fake news classification. The proposed model is based on the traditional inverse document frequency (TF-IDF) weighting measure. TF-IDF transforms the textual data into a weighted term-document matrix and is usually prone to the curse of dimensionality. Our novel statistical model filters the most discriminative terms to reduce the data's dimensionality and improve classification accuracy. We compare the proposed approach with the seven well-known feature selection and ranking techniques, namely normalized difference measure (NDM), binormal separation (BNS), odds ratio (OR), GINI, distinguished feature selector (DFS), information gain (IG), and Chi square (Chi). Our ensemble-based approach achieves high performance on two benchmark datasets, BET and UFN, achieving an accuracy of 88% and 90%, respectively. [ABSTRACT FROM AUTHOR]

A new depth-based clustering procedure for directional data is proposed. Such method is fully non-parametric and has the advantages to be flexible and applicable even in high dimensions when a suitable notion of depth is adopted. The introduced technique is evaluated through an extensive simulation study. In addition, a real data example in text mining is given to explain its effectiveness in comparison with other existing directional clustering algorithms. [ABSTRACT FROM AUTHOR]

Copyright of Contemporary Accounting Research is the property of Canadian Academic Accounting Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In modern China, industrialization has formed a critical foundation for the transition to modernization. However, the spatiotemporal evolution patterns and driving mechanisms of urban industrial development in Nanjing from 1840 to 1949 remain unclear. Based on textual historical sources, this study examined the spatiotemporal patterns of urban industrial development in Nanjing from 1840 to 1949 by using spatial analysis methods, GeoDetector, regression models and industrial structure indices. The results reveal the following: (1) The overall spatial distribution pattern of the industry in modern Nanjing exhibited a “one main, one secondary” dual-center “ladle-shaped” arrangement. Over time, industry has expanded from the urban center toward the east and north. (2) The modernization level of different industries was uneven, exhibiting a “center-periphery” spatial pattern. (3) At the micro level, transportation and population density were the primary influencing factors for industrial location, whereas at the macro level, government intervention mainly affected the industrialization pattern. (4) The industrial development pattern in modern Nanjing, in alignment with the “pole-axis” spatial system, serves as a microcosm of China’s urban modernization transition. This study represents the application of GIS methods in the humanities and provides valuable insights for urban planning and development.

Social media has become an essential source of news for everyday users. However, the rise of fake news on social media has made it more difficult for users to trust the information on these platforms. Most research studies focus on fake news detection in the English language, and only a limited number of studies deal with fake news in resource-poor languages such as Urdu. This article proposes a globally weighted term selection approach named normalized effect size (NES) to select highly discriminative features for Urdu fake news classification. The proposed model is based on the traditional inverse document frequency (TF-IDF) weighting measure. TF-IDF transforms the textual data into a weighted term-document matrix and is usually prone to the curse of dimensionality. Our novel statistical model filters the most discriminative terms to reduce the data’s dimensionality and improve classification accuracy. We compare the proposed approach with the seven well-known feature selection and ranking techniques, namely normalized difference measure (NDM), bi-normal separation (BNS), odds ratio (OR), GINI, distinguished feature selector (DFS), information gain (IG), and Chi square (Chi). Our ensemble-based approach achieves high performance on two benchmark datasets, BET and UFN, achieving an accuracy of 88% and 90%, respectively.

Copyright of Journal of Computer-Aided Design & Computer Graphics / Jisuanji Fuzhu Sheji Yu Tuxingxue Xuebao is the property of Gai Kan Bian Wei Hui and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Technological advancement is something we can no longer ignore, deep learning being one of those advancements. With deep learning, emotions can be extracted and analyzed in textual data to be utilized in various sectors, involving human behavior analysis. The main objective of this paper is to investigate and discover basic emotional categories that are typically exhibited by humans, such as sadness, joy, anger, fear, and disgust. Sectors focused on business can utilize emotion detection for various applications, such as to create personalized services, and in medical scenarios to help develop specialized mental treatments. This paper enables other researchers to enrich their knowledge and develop a more accurate and efficient deep learning model. To create an emotion detection model, analysis regarding the best performing model has to be done. This paper discusses the difference in performance across three different RNN models, namely LSTM, BiLSTM, and GRU. The RNN based models, in theory, share the same architecture and will be evaluated using the ISEAR dataset to create an emotion detection model. The result of this paper concluded that the simplest RNN model, GRU, achieved the highest score across four scoring metrics: accuracy, recall, precision, and F1 score. The GRU model achieved an accuracy of 60.26%, BiLSTM with 59.3%, and LSTM with 57.65%. Hence, based on the results, the GRU-based models are a strong choice for emotion detection when tested on the ISEAR dataset. [ABSTRACT FROM AUTHOR]

Predicting creditworthiness is an important task in the banking industry, as it allows banks to make informed lending decisions and manage risk. In this paper, we investigate the performance of two different deep learning credit scoring models developed on the textual descriptions of customer transactions available from open banking APIs. The first model is a deep learning model trained from scratch, while the second model uses transfer learning with a multilingual BERT model. We evaluate the predictive performance of these models using the area under the receiver operating characteristic curve (AUC) and Brier score. We find that a deep learning model trained from scratch outperforms a BERT transformer model finetuned on the same data. Furthermore, we find that SHAP can be used to explain such models both on a global level and for explaining rejections of actual applications. [ABSTRACT FROM AUTHOR]

Abstract Traffic flow prediction is an important part of an intelligent transportation system to alleviate congestion. In practice, most small and medium-sized activities are not given priority in transport planning, yet these activities often bring about a surge in demand for public transport. It is recognized that such patterns are inevitably more difficult to predict than those associated with day-to-day mobility, and that forecasting models built using traffic data alone are not comprehensive enough. Aiming at this problem, a depthwise separable convolutional fusion forecast network (FFN) was proposed by focusing on the impact of event information on traffic flow demand. FFN fused heterogeneous data to model traffic data, weather information, and event information extracted from the Internet. The depthwise separable one-dimensional convolution was used to encode the textual information describing the event layer by layer, and local one-dimensional sequence segments (ie subsequences) were extracted from the sequence to retain rich local semantic features. In the modeling process, the interaction of heterogeneous data was established, that is, the temporal and other data were used to drive the textual information representation in the encoding process to capture better relevant textual representations. Finally, information from different sources and formats was fused to obtain a joint feature representation tensor that predicts the traffic demand in the next day's event area. The experimental results show that the average absolute error of the fusion prediction network is reduced by 26.5%, the root mean square error is reduced by 11.6%, and the judgment coefficient is increased by 26.4% compared with the prediction network that only considers the traffic data.

Copyright of Iranian Journal of Insurance Research (IJIR) is the property of Iranian Journal of Insurance Research and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this study we explore the disciplinary diversity present within subject specialties in the social sciences and humanities. Subject specialties are operationalized as textually coherent clusters of documents. We apply topic modelling to textual information on the individual document level (titles and abstracts) to cluster a multilingual set of roughly 45,000 documents into subject specialties. The dataset includes the metadata of journal articles, conference proceedings, book chapters, and monographs. We make use of two indicators, namely, the organizational affiliation based on the departmental address of the authors and the cognitive orientation based on the disciplinary classifications at the publication level. First, we study the disciplinary diversity of the clusters by calculating a Hill-type diversity index. We draw an overall picture of the distribution of subject specialties over diversity scores and contrast the two indicators with each other. The goal is to discover whether some subject specialties are inherently multi- or interdisciplinary in nature, and whether the different indicators are telling a well-aligned, similar story. Second, for each cluster of documents we calculate the dominance, i.e. the relative size of the largest discipline. This proxy of disciplinary concentration gives an idea of the extent to which a specialty is disciplined. The results show that all subject specialties analyzed serve as interdisciplinary trading grounds, with outliers in both directions of the disciplinary-interdisciplinary continuum. For a large share of specialties, the dominant cognitive and organizational disciplinary classification were found to be well aligned. We present a typology of subject specialties by contrasting the organizational and cognitive diversity scores. [ABSTRACT FROM AUTHOR]

Smart health‐care is the innovation that leads to enhanced diagnostic tools, improved patient treatment, and gadgets that ease the quality of life for majority of people. Textual clinical documents about an individual contain sensitive and semantically corelated terms. Most privacy‐preserving approaches are not designed to prevent confidentiality threats. Although, recent approaches improved the utility of published output with generalized terms retrieved from several medical and general‐purpose knowledge bases like SNOMED‐CT and MASH. However, these models work on predefined sensitive terms using Wikipedia articles instead of authentic benchmarks. These Information Content‐based methods are not capable to achieve the best balance between privacy and utility. The existing approaches guarantee syntactic privacy by sanitization but lack semantic privacy for textual clinical data. Therefore, it is imperative to design a confidentiality‐aware framework to overcome these problems. Our proposed Confidentiality aware Textual Clinical Data Framework use preprocessed combinations of the terms instead of all combinations and perform automatic detection and sanitization of the sensitive and semantically correlated terms. The probabilistic sampling‐based method guarantees the semantic privacy. We use high‐level Petri nets to perform formal modeling of our proposed approach. Furthermore, we have also performed a detailed complexity analysis of the proposed framework. [ABSTRACT FROM AUTHOR]

I describe a brief summary of the development of databases used in accounting research and discuss the research questions addressed in traditional databases and ‘new’ databases. The new data include online searches such as Google Trends data; textual data from corporate disclosures, analyst reports, conference call transcripts, earnings press releases, and news media articles; social network and social media data from Twitter, LinkedIn, Glassdoor, and other data. New data holds promise for research on attention or cognitive processing constraints; on tone/valence, affect, deceptiveness and credibility for capital market and financial reporting outcomes. I examine the econometric challenges of new data and suggest the potential for new data to offer new auditing tools to detect poor financial reporting, which will help to discourage earnings management.

I describe a brief summary of the development of databases used in accounting research and discuss the research questions addressed in traditional databases and ‘new’ databases. The new data include online searches such as Google Trends data; textual data from corporate disclosures, analyst reports, conference call transcripts, earnings press releases, and news media articles; social network and social media data from Twitter, LinkedIn, Glassdoor, and other data. New data holds promise for research on attention or cognitive processing constraints; on tone/valence, affect, deceptiveness and credibility for capital market and financial reporting outcomes. I examine the econometric challenges of new data and suggest the potential for new data to offer new auditing tools to detect poor financial reporting, which will help to discourage earnings management.

O presente artigo pretende oferecer uma metodologia de pesquisa com o pacote Quanteda, que utiliza a linguagem R, aplicada à análise da obra de Fernando Pessoa. Quanteda (Quantitative Analysis of Textual Data) é um pacote de R para a manipulação e estudo de dados textuais. O programa objetiva aplicar processamento de linguagem natural a textos. Por sua vez, R é uma linguagem de programação para computação estatística suportada pelo R Core Team e R Foundation for Statistical Computing. A ferramenta, assim, permite o estudo textual quantitativo de corpus e oferece ferramentas de visualização que representam as análises. Desde topic modeling até redes semânticas ou análises de co-ocorrências, as ferramentas possibilitam estudos e representações detalhados de estruturas textuais.

This article aims to introduce Galileo, a platform for extracting and organizing news media data on social networks. Galileo integrates publications made on the main social networks used in the information ecosystem, namely Facebook, Twitter, and Instagram. Currently, the system includes 97 media outlets from nine countries: Brazil, Chile, Germany, Japan, Mexico, South Korea, Spain, United Kingdom, and United States. Galileo uses a Twitter API and the service Crowd-Tangle to download Facebook and Instagram posts. This data is stored in a local database and can be accessed through a user-friendly interface, which allows for the analysis of different characteristics of the posts, such as their text, source popularity, and temporal dimension. Galileo is a tool for researchers interested in understanding news cycles and analyzing news content on social networks. [ABSTRACT FROM AUTHOR]