Natural Language Processing Methods Research Articles

A case study of using natural language processing to extract consumer insights from tweets in American cities for public health crises

BackgroundThe COVID-19 pandemic was a “wake up” call for public health agencies. Often, these agencies are ill-prepared to communicate with target audiences clearly and effectively for community-level activations and safety operations. The obstacle is a lack of data-driven approaches to obtaining insights from local community stakeholders. Thus, this study suggests a focus on listening at local levels given the abundance of geo-marked data and presents a methodological solution to extracting consumer insights from unstructured text data for health communication.MethodsThis study demonstrates how to combine human and Natural Language Processing (NLP) machine analyses to reliably extract meaningful consumer insights from tweets about COVID and the vaccine. This case study employed Latent Dirichlet Allocation (LDA) topic modeling, Bidirectional Encoder Representations from Transformers (BERT) emotion analysis, and human textual analysis and examined 180,128 tweets scraped by Twitter Application Programming Interface’s (API) keyword function from January 2020 to June 2021. The samples came from four medium-sized American cities with larger populations of people of color.ResultsThe NLP method discovered four topic trends: “COVID Vaccines,” “Politics,” “Mitigation Measures,” and “Community/Local Issues,” and emotion changes over time. The human textual analysis profiled the discussions in the selected four markets to add some depth to our understanding of the uniqueness of the different challenges experienced.ConclusionsThis study ultimately demonstrates that our method used here could efficiently reduce a large amount of community feedback (e.g., tweets, social media data) by NLP and ensure contextualization and richness with human interpretation. Recommendations on communicating vaccination are offered based on the findings: (1) the strategic objective should be empowering the public; (2) the message should have local relevance; and, (3) communication needs to be timely.

A Comprehensive Survey of Deep Learning Techniques Natural Language Processing

In NLP research, unsupervised or semi-supervised learning techniques are increasingly getting more attention. These learning techniques are capable of learning from data that has not been manually annotated with the necessary answers or by combining non-annotated and annotated data. This essay presents a survey of various natural language processing methods. The discipline of natural language processing, which integrates linguistics, artificial intelligence, and computer science, was established to make it easier for computers and human language to communicate with one another. It is, as we can say, relevant psychopathology for the study of computer-human interaction. The understanding of natural language, which entails enabling machines to naturally interpret human language, is one of the many challenges this area faces. Discourse analysis, morphological separation, machine translation, production and understanding of NLP, part-of-speech tagging, recognition of optical characters, speech recognition, and sentiment analysis are some of the most frequent NLP tasks. As opposed to learning, which is supervised and typically yields few correct results for a given amount of input data, this job is typically quite difficult. However, there is a sizable amount of data available that is unannotated in nature, i.e. the entire contents are available on the internet, and it typically yields less accurate findings.

Natural Language Processing of Admission Notes to Predict Severe Maternal Morbidity During the Delivery Encounter

(Am J Obstet Gynecol. 2022;227:511.e1–511.e8) Natural language processing (NLP) is a machine learning technique that has been shown to enhance clinical prediction tasks with improved outcomes. This newer strategy is under investigation for its contribution to address severe maternal morbidity (SMM) beyond already in place predelivery planning, communication, and consultation. Using data inherently generated at a patient encounter, it is hypothesized that NLP may boost more accurate maternal risk stratification for SMM and allow provider alerts. NLP methods are proposed that stratify all patients into relevant risk pools to facilitate planning for appropriate care and potentially avert adverse outcomes.

Predicting Rupiah Sentiment Using Social Sentiment Analysis

In recent years, the behavior of foreign currency exchange rates (FOREX) has attracted a lot of attention from policy makers, investors, academics and regulators because due to its rapidly changing prices. FOREX predicting based on sentiment analysis has attracted a wide variety of research in finance and natural language processing. The availability of news, social media networks, and the rapid development of natural language processing methods result in better predictive performance. However, new studies related on Rupiah (Rupiah is legal money circulating in Indonesia) sentiment and its correlation with other FOREX are still rare and interesting to study.The purpose of this study is to (1) assess the accuracy of predicting Rupiah sentiment using social sentiment analysis and (2) to investigate the correlation between Rupiah sentiment and IDR currency exchange rate. The methodological approach used for this study is based on social sentiment analysis on Twitter and Pearson correlation.The result of this study is (1) from the seven models that we compared, the Decision Tree (DT) and Random Forest (RF) algorithm models have the best accuracy to avoid misclassification based on AUC. The accuracy measured by F1-score has values of 92.13% for both DT and RF, and (2) Rupiah sentiment has a strong positive correlation with USD (P-value: 0.017, r-value: 0.421) and SAR (P-value: 0.016, r-value: 0.419) exchange rate. On the other hand, Rupiah sentiment has a low significant negative correlation with AUD (P-value: 0.051, r-value: -0.403) and EUR (P-value: 0.079, r-value: -0.479). Rupiah sentiment also has a low but positive correlation with MYR (P-value: 0.073, r-value: 0.180) and is not significant with SGD (P-value: 0.119, r-value: -0.348) although it has a positive correlation. The strong positive correlation represents that the more positive Rupiah sentiment on Twitter, the stronger IDR exchange rate against another foreign currency, and vice versa.

Real or fictional? Digital entrepreneurial narratives and the acquisition of attentional resources in social entrepreneurship

Digital entrepreneurial narratives compensate for the limited interactivity of traditional entrepreneurship narratives. This is achieved by incorporating elements of social entrepreneurship into tangible, real or fictional scenarios, thereby providing immersive experiences for the audience. In this sense, social entrepreneurs construct shared meanings and form socially identifiable values through the joint participation, communication, and value creation of entrepreneurs and audiences. We conceptualized the digital entrepreneurial narrative and proposed documentary and fictional narratives as its most salient features. Then, we explored the effect of these narratives on attentional resources in a digital entrepreneurial narrative from the perspective of sensemaking theory. Based on novel methods of natural language processing and machine learning, the empirical results from an examination of 304 social entrepreneurs who participated in the short-video platforms indicated that both documentary and fictional narratives have a positive impact on the acquisition of attentional resources for social entrepreneurship. Additionally, we found the documentary aspect of digital entrepreneurial narratives attracts attentional resources by fostering a shared sense of meaningfulness. However, the mediating effect is less effective with fictional narratives. Our results broaden the study of entrepreneurial narratives, and also expand the content of social entrepreneurial legitimacy and sensemaking perspectives.

Conflict sensitive climate finance: lessons from the Green Climate Fund

ABSTRACT This study examines the state of integrated climate-security programming in the Green Climate Fund (GCF) and evaluates whether its operational activities and portfolio are conflict-sensitive and peace-responsive. Using a novel natural language processing method, the analysis draws on a comprehensive dataset of 1,704 documents published by the GCF from January 2012 to February 2023. The findings indicate that while the GCF adheres to conflict sensitivity principles, it falls short in implementing effective conflict governance practices. This oversight leads to the systematic underestimation of conflict risks, potentially exposing GCF projects to unforeseen operational challenges. On a positive note, the analysis also reveals signs of progress in integrated climate security programming in the GCF, primarily thanks to initiatives by the Board and Accredited Entities. Overall, this study offers novel insights into the work of the GCF that have potential practical implications for practitioners working in climate finance.

Public perception of waste regulations implementation. Natural language processing vs real GHG emission reduction modeling

The problem with waste generation and waste treatment that countries worldwide are facing, even after the implementation of measures, raises the question of the adequacy and viability of the regulations. The waste sector has been shown to contribute to a most notable increase in greenhouse gas (GHG) emissions in many countries worldwide, mainly coming from solid waste disposal and suggesting the need for a more profound observation of the waste management application. Therefore, this study employs an integrated approach to analyze the environmental, economic and social aspects of waste management, as the three fundamental pillars of sustainable development. Based on the Intergovernmental Panel on Climate Change (IPCC) methodology, a unique model was developed for the projection of GHG emissions in different scenarios, to evaluate the impact and cost of proposed waste policies and measures adopted to the reduction of greenhouse gas emissions, as well as the extent to which citizens themselves can contribute to the achievement of the goals. Furthermore, NLP (Natural Language Processing) methods were used to draw conclusions about the public opinion related to the waste sector, expressed by citizens and news media and measured on Twitter and news media teasers' data. The results show that more than 70% of the emissions can be reduced by 2050 with a significant contribution from the citizens, if all proposed measures are implemented, when compared to a scenario without policies and measures. The results also suggest a strong correlation between news reports and waste-related tweets, confirming an alignment in public perceptions and a response connected to waste management policies. The conclusions of this study offer valuable implications for policy-makers, suggesting the continuous observation and inclusion of citizens in the process of waste regulations implementation and effect.

Comparison between Machine Learning and Deep Learning Approaches for the Detection of Toxic Comments on Social Networks

The way we communicate has been revolutionised by the widespread use of social networks. Any kind of online message can reach anyone in the world almost instantly. The speed with which information spreads is undoubtedly the strength of social networks, but at the same time, any user of these platforms can see how toxic messages spread in parallel with likes, comments and ratings about any person or entity. In such cases, the victim feels even more helpless and defenceless as a result of the rapid spread. For this reason, we have implemented an automatic detector of toxic messages on social media. This allows us to stop toxicity in its tracks and protect victims. In particular, the aim of the survey is to demonstrate how traditional Machine Learning methods of Natural Language Processing (NLP) work on equal terms with Deep Learning methods represented by a Transformer architecture and characterised by a higher computational cost. In particular, the paper describes the results obtained by testing different supervised Machine Learning classifiers (Logistic Regression, Random Forest and Support Vector Machine) combined with two topic-modelling techniques of NLP, (Latent Semantic Analysis and Latent Dirichlet Allocation). A pre-trained Transformer named BERTweet was also tested. All models performed well in this task, so much so that values close to or above 90% were achieved in terms of the F1 score evaluation metric. The best result achieved by Transformer BERTweet, 91.40%, was therefore not impressive in this context, as the performance gains are too small compared to the computational overhead.

Exploring the Use of Natural Language Processing for Objective Assessment of Disorganized Speech in Schizophrenia.

Measurement-based care tools in psychiatry are useful for symptom monitoring and detecting response to treatment, but methods for quick and objective measurement are lacking especially for acute psychosis. The aim of this study was to explore potential language markers, detected by natural language processing (NLP) methods, as a means to objectively measure the severity of psychotic symptoms of schizophrenia in an acute clinical setting. Twenty-two speech samples were collected from seven participants who were hospitalized for schizophrenia, and their symptoms were evaluated over time with SAPS/SANS and TLC scales. Linguistic features were extracted from the speech data using machine learning techniques. Spearman's correlation was performed to examine the relationship between linguistic features and symptoms. Various machine learning models were evaluated by cross-validation methods for their ability to predict symptom severity using the linguistic markers. Reduced lexical richness and syntactic complexity were characteristic of negative symptoms, while lower content density and more repetitions in speech were predictors of positive symptoms. Machine learning models predicted severity of alogia, illogicality, poverty of speech, social inattentiveness, and TLC scores with up to 82% accuracy. Additionally, speech incoherence was quantifiable through language markers derived from NLP methods. These preliminary findings suggest that NLP may be useful in identifying clinically relevant language markers of schizophrenia, which can enhance objectivity in symptom monitoring during hospitalization. Further work is needed to replicate these findings in a larger data set and explore methods for feasible implementation in practice.

Trade vs. daily press: the role of news coverage and sentiment in real estate market analysis

ABSTRACT Each week, thousands of newspaper articles on real estate topics are read by market participants. While the market is comparatively intransparent, readers hope to find valuable information. This raises the question of whether this investment of time pays off and whether different types of newspapers are an equivalent source of information. This paper examines the relationship between news-based coverage of real estate topics respectively news-based market sentiment and total returns of the asset classes of residential, office and retail. Using methods of natural language processing, including word embedding, topic modelling and sentiment analysis, three sentiment indicators for each asset class can be derived from 137,000 articles of two trade and two daily newspapers. Our results suggest that trade newspapers outperform daily newspapers in the prediction of future total returns and that the generated sentiment indicators Granger-cause total returns. Moreover, the results indicate that daily newspapers report more negatively on rising returns in the residential market than the trade press. To the best knowledge of the authors, this is the first study to quantify news coverage and sentiment for the main real estate asset classes through means of textual analysis, and to assess different sentiments in trade and daily press.

Automatic extraction of social determinants of health from medical notes of chronic lower back pain patients.

We applied natural language processing and inference methods to extract social determinants of health (SDoH) information from clinical notes of patients with chronic low back pain (cLBP) to enhance future analyses of the associations between SDoH disparities and cLBP outcomes. Clinical notes for patients with cLBP were annotated for 7 SDoH domains, as well as depression, anxiety, and pain scores, resulting in 626 notes with at least one annotated entity for 364 patients. We used a 2-tier taxonomy with these 10 first-level classes (domains) and 52 second-level classes. We developed and validated named entity recognition (NER) systems based on both rule-based and machine learning approaches and validated an entailment model. Annotators achieved a high interrater agreement (Cohen's kappa of 95.3% at document level). A rule-based system (cTAKES), RoBERTa NER, and a hybrid model (combining rules and logistic regression) achieved performance of F1 = 47.1%, 84.4%, and 80.3%, respectively, for first-level classes. While the hybrid model had a lower F1 performance, it matched or outperformed RoBERTa NER model in terms of recall and had lower computational requirements. Applying an untuned RoBERTa entailment model, we detected many challenging wordings missed by NER systems. Still, the entailment model may be sensitive to hypothesis wording. This study developed a corpus of annotated clinical notes covering a broad spectrum of SDoH classes. This corpus provides a basis for training machine learning models and serves as a benchmark for predictive models for NER for SDoH and knowledge extraction from clinical texts.

The role of occupational skill sets in the digital transformation: how IT progress shapes returns to specialization and social skills

Workers’ occupational skill sets play a crucial role in successfully handling digital transformation. We investigate whether and how different types of occupational skill sets benefit from digital transformation. We theoretically and empirically analyze wage returns of workers in occupations with more or less specialized skill sets and with more or less social skills when IT increases in their industry. Applying natural language processing methods to the texts of occupational training curricula, we develop measures for occupational specialization and social skills. We use vocational education and training curricula from Switzerland because they cover approx. two-thirds of the working population. Using curricula, industry-level IT data and individual-level administrative wage data, our individual fixed-effects analyses show that IT progress leads to higher wage returns for workers in highly specialized occupations but not for workers in more general occupations. In addition, we find that high levels of social skills cannot make up for this difference when IT advances. However, our results indicate that for workers with high specialization, a combination with high social skills generates additional benefits when IT advances. Overall, our results suggest that, contrary to typical assumptions in educational policy debates, workers with specialized occupational skill sets—possibly in combination with high social skills—appear to be the ones who are particularly well prepared to cope with digital transformation.

OC-019 LEVERAGING NATURAL LANGUAGE PROCESSING AND ARTIFICIAL INTELLIGENCE TO LABEL UNSTRUCTURED DATA FOR RISK PREDICTION

Abstract Aim Majority of clinical data is categorized into unstructured data, including operative notes containing unexplored key details yet to be incorporated into a surgical risk model. We sought to determine whether Natural Language Processing (NLP) methods could be implemented to systematically capture operative note features that can be considered for patient-specific risk modeling. Methods Operative notes describing abdominal wall closure for 8233 patients undergoing colorectal operations between 2013–2019 were obtained. A development dataset was manually generated for 600 patients including skin and fascia suture material, suture size, and closure pattern. These features were analyzed to train/test various NLP approaches and assess their corresponding accuracy. Results Of the remaining 7633 operative notes, assessment of suture type, size, and closure pattern for skin and fascia was analyzed by keyword search, regular expression, and RoBERTa 5-fold Cross Validation(CV). Overall, RoBERTa 5-fold CV demonstrated superiority in all feature extraction tasks. For fascial suture type, RoBERTa 5-fold CV achieved 92.1% accuracy, regular expression achieved 83.5%, and keyword search achieved 9.2%. For fascial suture size, RoBERTa 5-fold CV achieved 91.6% accuracy, regular expression achieved 89.4%, and keyword search achieved 81%. For fascial closure pattern, RoBERTa 5-fold CV achieved 91.4% accuracy, regular expression achieved 80.8%, and keyword search achieved 24.7%. Skin feature analysis also demonstrated superiority of RoBERTa 5-fold CV while assessing suture type, size, and closure pattern with 93.3%, 93.3%, and 84.9% accuracy, respectively. Conclusions NLP methods can be reliably used to label key features contained within unstructured datasets, enabling use in predictive modeling in medicine.

Understanding Views Around the Creation of a Consented, Donated Databank of Clinical Free Text to Develop and Train Natural Language Processing Models for Research: Focus Group Interviews With Stakeholders

Information stored within electronic health records is often recorded as unstructured text. Special computerized natural language processing (NLP) tools are needed to process this text; however, complex governance arrangements make such data in the National Health Service hard to access, and therefore, it is difficult to use for research in improving NLP methods. The creation of a donated databank of clinical free text could provide an important opportunity for researchers to develop NLP methods and tools and may circumvent delays in accessing the data needed to train the models. However, to date, there has been little or no engagement with stakeholders on the acceptability and design considerations of establishing a free-text databank for this purpose. This study aimed to ascertain stakeholder views around the creation of a consented, donated databank of clinical free text to help create, train, and evaluate NLP for clinical research and to inform the potential next steps for adopting a partner-led approach to establish a national, funded databank of free text for use by the research community. Web-based in-depth focus group interviews were conducted with 4 stakeholder groups (patients and members of the public, clinicians, information governance leads and research ethics members, and NLP researchers). All stakeholder groups were strongly in favor of the databank and saw great value in creating an environment where NLP tools can be tested and trained to improve their accuracy. Participants highlighted a range of complex issues for consideration as the databank is developed, including communicating the intended purpose, the approach to access and safeguarding the data, who should have access, and how to fund the databank. Participants recommended that a small-scale, gradual approach be adopted to start to gather donations and encouraged further engagement with stakeholders to develop a road map and set of standards for the databank. These findings provide a clear mandate to begin developing the databank and a framework for stakeholder expectations, which we would aim to meet with the databank delivery.

Fake Review Detection System: A Review

Fake reviews, often referred to as deceptive or dishonest reviews, have become a significant concern for both businesses and consumers (Feng et al., 2016). These reviews are deliberately crafted to mislead or manipulate the opinions of others and can be driven by various motives, such as financial gain, competition, or personal grudges (Liu et al., 2018). The impact of fake reviews on the reputation and sales of products or services can be substantial, leading consumers to make misguided purchasing decisions (Wang et al., 2017). The demand for effective methods to detect and address fake reviews has been steadily increasing, aiming to assist businesses and consumers in identifying and mitigating the influence of such reviews (Hu et al., 2019). However, detecting fake reviews presents challenges due to the diverse forms they can take and the difficulty in distinguishing them from genuine reviews (Xu et al., 2020). To tackle this challenge, researchers have developed various approaches and techniques, including machine learning, natural language processing, crowd-sourced annotation, and content-based methods (Feng et al., 2016; Liu et al., 2018). This review provides a comprehensive overview of the current state of the art in fake review detection methods, highlighting the key challenges and limitations associated with these approaches. The implications of these methods for businesses and consumers are also discussed, and future research directions are suggested to further enhance the effectiveness and reliability of fake review detection techniques.

Determining psycholinguistic features of deception in phishing messages

PurposeDistinguishing phishing emails from legitimate emails continues to be a difficult task for most individuals. This study aims to investigate the psycholinguistic factors associated with deception in phishing email text and their effect on end-user ability to discriminate phishing emails from legitimate emails.Design/methodology/approachEmail messages and end-user decisions collected from a laboratory phishing study were validated and analyzed using natural language processing methods (Linguistic Inquiry Word Count) and penalized regression models (LASSO and Elastic Net) to determine the linguistic dimensions that attackers may use in phishing emails to deceive end-users and measure the impact of such choices on end-user susceptibility to phishing.FindingsWe found that most participants, who played the role of a phisher in the study, chose to deceive their end-user targets by pretending to be a familiar individual and presenting time pressure or deadlines. Results show that use of words conveying certainty (e.g. always, never) and work-related features in the phishing messages predicted higher end-user vulnerability. On the contrary, use of words that convey achievement (e.g. earn, win) or reward (cash, money) in the phishing messages predicted lower end-user vulnerability because such features are usually observed in scam-like messages.Practical implicationsInsights from this research show that analyzing emails for psycholinguistic features associated with computer-mediated deception could be used to fine-tune and improve spam and phishing detection technologies. This research also informs the kinds of phishing attacks that must be prioritized in antiphishing training programs.Originality/valueApplying natural language processing and statistical modeling methods to analyze results from a laboratory phishing experiment to understand deception from both attacker and end-user is novel. Furthermore, results from this work advance our understanding of the linguistic factors associated with deception in phishing email text and its impact on end-user susceptibility.

Translating Akkadian to English with neural machine translation.

Cuneiform is one of the earliest writing systems in recorded human history (ca. 3,400 BCE-75 CE). Hundreds of thousands of such texts were found over the last two centuries, most of which are written in Sumerian and Akkadian. We show the high potential in assisting scholars and interested laypeople alike, by using natural language processing (NLP) methods such as convolutional neural networks (CNN), to automatically translate Akkadian from cuneiform Unicode glyphs directly to English (C2E) and from transliteration to English (T2E). We show that high-quality translations can be obtained when translating directly from cuneiform to English, as we get 36.52 and 37.47 Best Bilingual Evaluation Understudy 4 (BLEU4) scores for C2E and T2E, respectively. For C2E, our model is better than the translation memory baseline in 9.43, and for T2E, the difference is even higher and stands at 13.96. The model achieves best results in short- and medium-length sentences (c. 118 or less characters). As the number of digitized texts grows, the model can be improved by further training as part of a human-in-the-loop system which corrects the results.

Empirical Validation of Automated Vulnerability Curation and Characterization

Prior research has shown that public vulnerability systems such as US National Vulnerability Database (NVD) rely on a manual, time-consuming, and error-prone process which has led to inconsistencies and delays in releasing final vulnerability results. This work provides an approach to curate vulnerability reports in real-time and map textual vulnerability reports to machine readable structured vulnerability attribute data. Designed to support the time consuming human analysis done by vulnerability databases, the system leverages the Common Vulnerabilities and Exposures (CVE) list of vulnerabilities and the vulnerability attributes described by the National Institute of Standards and Technology (NIST) Vulnerability Description Ontology (VDO) framework. Our work uses Natural Language Processing (NLP), Machine Learning (ML) and novel Information Theoretical (IT) methods to provide automated techniques for near real-time publishing, and characterization of vulnerabilities using 28 attributes in 5 domains. Experiment results indicate that vulnerabilities can be evaluated up to 95 hours earlier than using manual methods, they can be characterized with F-Measure values over 0.9, and the proposed automated approach could save up to 47% of the time spent for CVE characterization.

Natural Language Processing Methods to Empirically Explore Social Contexts and Needs in Cancer Patient Notes.

There is an unmet need to empirically explore and understand drivers of cancer disparities, particularly social determinants of health. We explored natural language processing methods to automatically and empirically extract clinical documentation of social contexts and needs that may underlie disparities. This was a retrospective analysis of 230,325 clinical notes from 5,285 patients treated with radiotherapy from 2007 to 2019. We compared linguistic features among White versus non-White, low-income insurance versus other insurance, and male versus female patients' notes. Log odds ratios with an informative Dirichlet prior were calculated to compare words over-represented in each group. A variational autoencoder topic model was applied, and topic probability was compared between groups. The presence of machine-learnable bias was explored by developing statistical and neural demographic group classifiers. Terms associated with varied social contexts and needs were identified for all demographic group comparisons. For example, notes of non-White and low-income insurance patients were over-represented with terms associated with housing and transportation, whereas notes of White and other insurance patients were over-represented with terms related to physical activity. Topic models identified a social history topic, and topic probability varied significantly between the demographic group comparisons. Classification models performed poorly at classifying notes of non-White and low-income insurance patients (F1 of 0.30 and 0.23, respectively). Exploration of linguistic differences in clinical notes between patients of different race/ethnicity, insurance status, and sex identified social contexts and needs in patients with cancer and revealed high-level differences in notes. Future work is needed to validate whether these findings may play a role in cancer disparities.

Создание средств интеллектуальной поддержки процедур рекрутинга инновационного предприятия на основе методов обработки естественного языка

When conducting recruiting procedures for a modern innovative company, the task of creating automated solutions designed to objectively determine the competencies of an employee is most urgent. There have been de-scribed the results of an analytical review of methods that allow identifying the competencies of specialists based on semantic studies of texts generated by them, followed by analysis of the applicability of such solutions to perform specific recruiting procedures. There is given a description of an approach to creating collections of texts characterizing the competencies of the applicant for the vacancy, which are classified according to the time of text generation (retrospective experience /current activity) and interaction with co-authors. There have been considered the examples of using different methods of Natural language processing based on the vector representation of texts for solving specific tasks related to the assessment of the competencies of specialists and the tools for semantic analysis of texts.