Text Mining Model Research Articles

Tasting the Port wine cellar experience: what features please the most?

ABSTRACT This study addresses wine tasting experienced by tourists in the context of the worldwide renowned Port wine. A total of 824 online reviews posted about four major Port wine cellars in TripAdvisor were collected. Data were analyzed with text mining and topic modeling, and sentiment scores, as well as ratings, were also considered. Overall, results showed that guests were primarily attracted by the experience as a whole, including the tour and the wine culture and not just the wine tasting experience. Furthermore, six topics were found with associated main keywords and comprehending distinct groups of guests with different preferences. Three groups underlined tour and visiting experience, whereas two other groups were composed of enthusiastic wine fans, with one more focused on the wine/winery/cellar experience while the other enjoyed a diversified wine selection. Yet another group aggregated tourists who valued the food experience.

Comparison of Text Mining Models for Food and Dietary Constituent Named-Entity Recognition

Biomedical Named-Entity Recognition (BioNER) has become an essential part of text mining due to the continuously increasing digital archives of biological and medical articles. While there are many well-performing BioNER tools for entities such as genes, proteins, diseases or species, there is very little research into food and dietary constituent named-entity recognition. For this reason, in this paper, we study seven BioNER models for food and dietary constituents recognition. Specifically, we study a dictionary-based model, a conditional random fields (CRF) model and a new hybrid model, called FooDCoNER (Food and Dietary Constituents Named-Entity Recognition), which we introduce combining the former two models. In addition, we study deep language models including BERT, BioBERT, RoBERTa and ELECTRA. As a result, we find that FooDCoNER does not only lead to the overall best results, comparable with the deep language models, but FooDCoNER is also much more efficient with respect to run time and sample size requirements of the training data. The latter has been identified via the study of learning curves. Overall, our results not only provide a new tool for food and dietary constituent NER but also shed light on the difference between classical machine learning models and recent deep language models.

Analysing perceptions towards electric cars using text mining and sentiment analysis: A case study of the newly introduced TOGG in Turkey

The electric car market is growing steadily around the world and, accordingly, has become an attractive research area. It is important to understand the consumer perspective on newly introduced electric cars, such as those of Turkey’s Automobile Joint Venture Group Inc. (TOGG). Thus, the purpose of this study is to provide a better understanding of consumers’ perceptions related to the newly introduced TOGG, which may create a competitive advantage. Social media is an abundant source of textual data that allows for very reliable analysis and understanding of consumer opinions. In this study, Twitter comments on TOGG were collected and studied. Text mining, sentiment analysis and topic model analysis were then conducted. The results show that TOGG is a popular product with the public: there are many more positive Twitter comments related to TOGG than negative ones. The topics identified in social media are price expectancy, production facility, design and features. The most frequent topic for both positive and negative comments is price expectancy.

텍스트 마이닝과 토픽 모델링 기법을 활용한 노인복지주택의 국내연구 경향 분석

The purpose of this study is to analyze domestic research papers published since 1981 in order to identify the research trend of welfare housing for the elderly. The research topic, research purpose, and research method of the papers were exploring, and then, the keywords of the papers were analyzed using text mining and LDA topic modeling. The results are summarized as follows. Firstly, papers related to welfare housing for the elderly increased quantitatively according to the enactment of policies and laws, but recently decreased. On the other hand, research fields, research topics, and research methods are gradually diversifying, and multidisciplinary studies are being integrated. Secondly, by analyzing the frequency and degree centrality of the keywords through text mining, it was possible to identify the topics that were frequently dealt with and the subjects that were treated as important. In addition, through the network analysis between the keywords, the connection structure and strength of related to the subject, purpose, and method of the study could be grasped. Thirdly, as a result of the topic modeling analysis, it can be seen that important research topics are

뉴스 텍스트를 활용한 국내 자동차 시장의 브랜드 속성 기반 시장 구조 분석

We propose an approach to explore the discourse generated by the public press to draw meaningful insights on brands, brand attributes, and the brand attributes-based market structure. Focusing on the Korean automobile market, we collect 346,795 news articles related to the automobile industry. Then, we employ a machine learning-based text mining model to extract the semantic structure of brands and brand attributes reflected in the news text data. Specifically, we utilize the Word2vec model to assign each word as a vector, and measure the semantic (dis)similarity as the vector distance between word vectors. This approach enables us to find key brand attributes in the Korean automobile market, such as “quality”, “luxury”, and “trust”, and to measure the strength of their associations with automobile brands. Using the associations between brand and brand attributes, we visualize the brand-level market structure via perceptual maps that reveal the competitive nature of the market at the brand level. Based on these results, we summarize meaningful insights for the brand management and marketing strategy.

Multi-criteria text mining model for COVID-19 testing reasons and symptoms and temporal predictive model for COVID-19 test results in rural communities.

This study is conducted to build a multi-criteria text mining model for COVID-19 testing reasons and symptoms. The model is integrated with a temporal predictive classification model for COVID-19 test results in rural underserved areas. A dataset of 6895 testing appointments and 14 features is used in this study. The text mining model classifies the notes related to the testing reasons and reported symptoms into one or more categories using look-up wordlists and a multi-criteria mapping process. The model converts an unstructured feature to a categorical feature that is used in building the temporal predictive classification model for COVID-19 test results and conducting some population analytics. The classification model is a temporal model (ordered and indexed by testing date) that uses machine learning classifiers to predict test results that are either positive or negative. Two types of classifiers and performance measures that include balanced and regular methods are used: (1) balanced random forest and (2) balanced bagged decision tree. The balanced or weighted methods are used to address and account for the biased and imbalanced dataset and to ensure correct detection of patients with COVID-19 (minority class). The model is tested in two stages using validation and testing sets to ensure robustness and reliability. The balanced classifiers outperformed regular classifiers using the balanced performance measures (balanced accuracy and G-score), which means the balanced classifiers are better at detecting patients with positive COVID-19 results. The balanced random forest achieved the best average balanced accuracy (86.1%) and G-score (86.1%) using the validation set. The balanced bagged decision tree achieved the best average balanced accuracy (83.0%) and G-score (82.8%) using the testing set. Also, it was found that the patient history, age, testing reasons, and time are the key features to classify the testing results.

Corporate Disclosures in Pandemic Times. The Annual and Interim Reports Case

ABSTRACT The paper examines if COVID-19 crisis has brought changes in companies’ approach on corporate reporting, with focus on annual reports. The research method is based on text mining techniques in order to build measures of readability and tone of uncertainty of annual reports, given the information published by the companies listed on four stock exchanges from Europe, namely the Bucharest Stock Exchange, ATHEX Stock Exchange, IBEX-35, and WIG-20 between 2017–2020. Findings emphasize, through text mining, multivariate analysis, and topic modeling, that the analyzed reports are less extensive in times of pandemic and tend to become more generic. Among firms’ financial performance metrics considered in our models, we found that there is a significant association only between annual reports textual characteristics and respectively, firm size, price earnings ratio and accruals reported. We prove as well significant stock exchange effects and industry effects. Our results show a slight decrease in annual reports readability, while the tone of uncertainty is more prominent within firms listed on less mature stock exchanges.

Applying Automated Originality Scoring to the Verbal Form of Torrance Tests of Creative Thinking

In this study, we applied different text-mining methods to the originality scoring of the Unusual Uses Test (UUT) and Just Suppose Test (JST) from the Torrance Tests of Creative Thinking (TTCT)–Verbal. Responses from 102 and 123 participants who completed Form A and Form B, respectively, were scored using three different text-mining methods. The validity of these scoring methods was tested against TTCT’s manual-based scoring and a subjective snapshot scoring method. Results indicated that text-mining systems are applicable to both UUT and JST items across both forms and students’ performance on those items can predict total originality and creativity scores across all six tasks in the TTCT-Verbal. Comparatively, the text-mining methods worked better for UUT than JST. Of the three text-mining models we tested, the Global Vectors for Word Representation (GLoVe) model produced the most reliable and valid scores. These findings indicate that creativity assessment can be done quickly and at a lower cost using text-mining approaches.

Identification of risk features using text mining and BERT-based models: Application to an oil refinery

The uncontrollable release of hazardous substances may lead to catastrophic accidents. In this context, risk studies are aimed at recommending either preventive measures or designing safeguards to mitigate the consequences. To that end, risk experts postulate possible leakages, then identify their causes and consequences, and finally evaluate and classify the risks into categories. These analyses rely on examination different engineering textual documents and attendance numerous meetings, which is very time consuming. Moreover, this qualitative process of hazard identification and assessment are usually the first steps of quantitative risk analysis (QRA) and is paramount to ensure its quality. Therefore, we here propose to use text mining and fine-tuned trained bidirectional encoder representations from transformers (BERT) models to support and reduce the efforts required for completing the early stages of QRA. Our idea is to apply these techniques to identify the potential consequences of accidents related to the operation of an oil refinery and classify each scenario in terms of severity of the consequence and likelihood of occurrence. The proposed method was applied to an actual oil refinery and presented very promising results. The potential consequences, the severity and likelihood categories were predicted with a mean accuracy of 97.42%, 86.44%, and 94.34% respectively. The models resulting from this research were embedded into a web-based app that is called HALO (hazard analysis based on language processing for oil refineries).

Automatic consistency assurance for literature-based gene ontology annotation

BackgroundLiterature-based gene ontology (GO) annotation is a process where expert curators use uniform expressions to describe gene functions reported in research papers, creating computable representations of information about biological systems. Manual assurance of consistency between GO annotations and the associated evidence texts identified by expert curators is reliable but time-consuming, and is infeasible in the context of rapidly growing biological literature. A key challenge is maintaining consistency of existing GO annotations as new studies are published and the GO vocabulary is updated.ResultsIn this work, we introduce a formalisation of biological database annotation inconsistencies, identifying four distinct types of inconsistency. We propose a novel and efficient method using state-of-the-art text mining models to automatically distinguish between consistent GO annotation and the different types of inconsistent GO annotation. We evaluate this method using a synthetic dataset generated by directed manipulation of instances in an existing corpus, BC4GO. We provide detailed error analysis for demonstrating that the method achieves high precision on more confident predictions.ConclusionsTwo models built using our method for distinct annotation consistency identification tasks achieved high precision and were robust to updates in the GO vocabulary. Our approach demonstrates clear value for human-in-the-loop curation scenarios.

Measuring divergent thinking originality with human raters and text-mining models: A psychometric comparison of methods.

Measuring divergent thinking originality with human raters and text-mining models: A psychometric comparison of methods.

BioVAE: a pre-trained latent variable language model for biomedical text mining.

SummaryLarge-scale pre-trained language models (PLMs) have advanced state-of-the-art (SOTA) performance on various biomedical text mining tasks. The power of such PLMs can be combined with the advantages of deep generative models. These are examples of these combinations. However, they are trained only on general domain text, and biomedical models are still missing. In this work, we describe BioVAE, the first large-scale pre-trained latent variable language model for the biomedical domain, which uses the OPTIMUS framework to train on large volumes of biomedical text. The model shows SOTA performance on several biomedical text mining tasks when compared to existing publicly available biomedical PLMs. In addition, our model can generate more accurate biomedical sentences than the original OPTIMUS output.Availability and implementationOur source code and pre-trained models are freely available: https://github.com/aistairc/BioVAE.Supplementary information Supplementary data are available at Bioinformatics online.

Iranian COVID-19 Publications in LitCovid: Text Mining and Topic Modeling

COVID-19 is a threat to the lives of people all over the world. As a result of the new and unknown nature of COVID-19, much research has been conducted recently. In order to increase and enhance the growth rate of Iranian publications on COVID-19, this article aims to analyze these publications in LitCovid to identify the topical and content structure and topic modeling of scientific publications in the mentioned subject area. The present article is applied research performed by using an analytical approach as well as text mining techniques. The statistical population is all the publications of Iranian researchers in LitCovid. Latent Dirichlet Allocation (LDA) and Python were used to analyze the data and implement text mining and topic modeling algorithms. Data analysis shows that the percentage of Iranian publications in the eight topical groups in LitCovid is as follows: prevention (39.57%), treatment (18.99%), diagnosis (18.99%), forecasting (7.83%), case report (6.52%), mechanism (3.91%), transmission (3.62%), and general (0.58%). The results indicate that patient, pandemic, outbreak, case, Iranian, model, care, health, coronavirus, and disease are the most important words in the publications of Iranian researchers in LitCovid. Six topics for prevention; four topics for treatment and case report and forecasting; three topics for diagnosis, mechanism, and transmission in general have been obtained by implementing the topic modeling algorithm. Most of the Iranian publications in LitCovid are related to the topic “pandemic status,” with 22.47% in the prevention category, and the lowest number of publications is related to the topic “environment,” with 11.11% in the transmission category. The present study indicates a better understanding of essential and strategic issues of Iranian publications in LitCovid. The results reveal that many Iranian studies on COVID-19 were primarily on the issues related to prevention, management, and control. These findings provided a structured and research-based viewpoint of COVID-19 in Iran to guide researchers and policymakers.

Challenging suicide, burnout, and depression among veterinary practitioners and students: text mining and topics modelling analysis of the scientific literature

BackgroundWorldwide, veterinary practitioners and students are reported to be at higher risk of suicide, burnout, and depression compared to other occupational groups. The aim of the current study was to apply text mining and topic modelling analysis on scientific literature regarding suicide, burnout, and depression among veterinary practitioners and students to extract meaningful and synthetic information. These statistical approaches can be used to comprehend more in deep the phenomena involving veterinarians and veterinary students and to suggest the potential changes needed in admission to veterinary school, veterinary curricula, and post-graduation initiatives as preventive actions.ResultsA systematic search protocol was set up to identify scientific literature that published on the topic from 1985 to 2019. Two-hundred-eleven records were selected with abstracts/texts submitted to text mining and topic modelling analysis. Student, stress, work, anim*, and euthanasia resulted the most frequent terms. Topics modelling allowed to differentiate groups of words and papers in 3 areas of interest: 1) students’ difficulties encountered during their studies that increase stress and anxiety impairing their psychological health; 2) exposure to death and euthanasia as risk factor for mental health; and 3) need of support among those providing medical and health care, and of supportive group work to cope with such profession.ConclusionBased on the most frequent words included in the clouds and on the contents of the papers clusterised in them, some suggestions are interfered. It is emphasized that the veterinary curricula should include courses that prepare them early to deal with animal death and post-death grief of pet owners, to handle ethical dilemmas and moral stressors, to communicate with clients and staff members, to work in team, to balance work-family life and to promote individual and team resources. Specific courses for veterinary practitioners could keep them updated on their new roles and ways to handle them among functioning as potential feedbacks to monitor their psychological wellbeing.

Impacts of COVID-19 on Transport Modes and Mobility Behavior: Analysis of Public Discourse in Twitter.

This study proposes a framework to analyze public discourse in Twitter to understand the impacts of COVID-19 on transport modes and mobility behavior. It also identifies reopening challenges and potential reopening strategies that are discussed by the public. First, the study collects 15,776 tweets that relate to personal opinions on transportation services posted between May 15 and June 15, 2020. Next, it applies text mining and topic modeling techniques to the tweets to determine the prominent themes, terms, and topics in those discussions to understand public feelings, behavior, and broader sentiments about the changes brought about by COVID-19 on transportation systems. Results reveal that people are avoiding public transport and shifting to using private car, bicycle, or walking. Bicycle sales have increased remarkably but car sales have declined. Cycling and walking, telecommuting, and online schools are identified as possible solutions to COVID-19 mobility problems and to reduce car usage with an aim to tackle traffic congestion in the post-pandemic world. People appreciated government decisions for funding allocation to public transport, and asked for the reshaping, restoring, and safe reopening of transit systems. Protecting transit workers, riders, shop customers and staff, and office employees is identified as a crucial reopening challenge, whereas mask wearing, phased reopening, and social distancing are proposed as effective reopening strategies. This framework can be used as a tool by decision makers to enable a holistic understanding of public opinions on transportation services during COVID-19 and formulate policies for a safe reopening.

From intuition to intelligence: a text mining–based approach for movies' green-lighting process

PurposeThe purpose of this paper is to develop a predictive model for box office performance based on the textual information in movie scripts in the green-lighting process of movie production.Design/methodology/approachThe authors use Latent Dirichlet Allocation to determine the hidden textual structure in movie scripts by extracting topic probabilities as predictors for classification. The extracted topic probabilities are used as inputs for the predictive model for the box office performance. For the predictive model, the authors utilize a variety of classification algorithms such as logistic classification, decision trees, random forests, k-nearest neighbor algorithms, support vector machines and artificial neural networks, and compare their relative performances in predicting movies' market performance.FindingsThis approach for extracting textual information from movie scripts produces a valuable typology for movies. Moreover, our modeling approach has significant power to predict movie scripts' profitability. It provides a superior prediction performance compared to previous benchmarks, such as that of Eliashberg et al. (2007).Research limitations/implicationsThis work contributes to literature on predicting the box office performance in the green-lighting process and literature regarding suggesting models for the idea screening stage in the new product development process. Besides, this is one of the few studies that use movie script data to predict movies' financial performance by proposing an approach to integrate text mining models and machine learning algorithms with movie experts' intuition.Practical implicationsFirst, the authors’ approach can significantly reduce the financial risk associated with movie production decisions before the pre-production stage. Second, this paper proposes an approach that is applicable at a very early stage of new product development, such as the idea screening stage. The authors also introduce an online-based movie scenario database system that can help movie studios make more systematic and profitable decisions in the green-lighting process. Third, this approach can help movie studios estimate movie scripts' financial value.Originality/valueThis study is one of the few studies to forecast market performance in the green-lighting process.

The functional roles of metacognitive judgement and emotion in predicting clinical reasoning performance with a computer simulated environment

ABSTRACT Monitoring one’s learning activities is a key component of self-regulated learning (SRL) leading to successful learning and performance outcomes across settings. Achievement emotions also play an important part in SRL and consequently student learning outcomes. However, there is little research on how specific types of monitoring (i.e. Feeling-of-Knowing, Judgement-of-Learning) and achievement emotions jointly affect performance in clinical reasoning within computer simulated environments. In this study, we explored the joint predictive value of monitoring judgements and achievement emotions in medical students’ ability to make accurate and efficient clinical diagnoses using BioWorld, a computer-based learning environment designed to help medical students practice clinical reasoning skills. Multiple types of data (log files, verbal protocols, self-report questionnaires) from 27 students were analyzed using text mining and linear mixed-effects models. We found that feeling-of-knowing judgements positively predicted diagnostic efficiency whereas judgement-of-learning and the achievement emotion of anger negatively predicted diagnostic efficiency. Achievement emotion of pride positively predicted confidence in diagnosing the case correctly. This study not only provides theoretical and methodological insights but also alert medical instructors to the many dimensions of diagnostic performances and effective intervention strategies in clinical reasoning.

A Domain-Based Model for Knowledge Acquisition and Text Mining in A Database

A Domain-Based Model for Knowledge Acquisition and Text Mining in A Database

Research on Judicial Big Data Text Mining and Sentencing Prediction Model

In the era of big data, computer data has grown tremendously, especially in unstructured text data, and its growth rate is amazing. For this reason, in the application of judicial big data, people need to apply automated text mining technology to process unstructured data and predict the specific value of information. This article summarizes the research design and technical background content based on previous work experience. The author discusses the analysis of specific research results from three aspects: pre-processing of research data, establishment of sentencing prediction models, model testing and evaluation.

What do economic education scholars study? Insights from machine learning

The authors of this article use text mining techniques to uncover hidden or latent topics in economic education. The common use of JEL codes only identifies the academic setting for each paper but does not identify the underlying economic concept the paper addresses. An unsupervised machine learning algorithm called Latent Dirichlet Allocation is utilized to identify 15 hidden topics in economic education scholarly work. The text mining model identifies economic education topics by finding correlations in word usage across different documents. The authors show that these newly identified research topics explain more variation in citation counts than the commonly adopted JEL codes. Moreover, specific journals display preferences for certain topics within economic education research.