Social Network Posts Research Articles

Federated learning-based natural language processing: a systematic literature review

Federated learning (FL) is a decentralized machine learning (ML) framework that allows models to be trained without sharing the participants’ local data. FL thus preserves privacy better than centralized machine learning. Since textual data (such as clinical records, posts in social networks, or search queries) often contain personal information, many natural language processing (NLP) tasks dealing with such data have shifted from the centralized to the FL setting. However, FL is not free from issues, including convergence and security vulnerabilities (due to unreliable or poisoned data introduced into the model), communication and computation bottlenecks, and even privacy attacks orchestrated by honest-but-curious servers. In this paper, we present a systematic literature review (SLR) of NLP applications in FL with a special focus on FL issues and the solutions proposed so far. Our review surveys 36 recent papers published in relevant venues, which are systematically analyzed and compared from multiple perspectives. As a result of the survey, we also identify the most outstanding challenges in the area.

Entity Linking for real-time geolocation of natural disasters from social network posts.

When a fast kinetic natural disaster occurs, it is crucial that crisis managers quickly understand the extent of the situation, especially through the development of "big picture" maps. For many years, great efforts have been made to use social networks to help build this situational awareness. While there are many models for automatically extracting information from posts, the difficulty remains in detecting and geolocating this information on the fly so that it can be placed on maps. Whilst most of the work carried out to date on this subject has been based on data in English, we tackle the problem of detecting and geolocating natural disasters from French messages posted on the Twitter platform (now renamed "X"). To this end, we first build an appropriate dataset comprised of documents from the French Wikipedia corpus, the dataset from the CAp 2017 challenge, and a homemade annotated Twitter dataset extracted during French natural disasters. We then developed an Entity-Linking pipeline in adequacy with our end-application use case: real-time prediction and peak resiliency. We show that despite these two additional constraints, our system's performances are on par with state-of-the-art systems. Moreover, the entities geolocated by our model show a strong coherence with the spatiotemporal signature of the natural disasters considered, which suggests that it could usefully contribute to automatic social network analysis for crisis managers.

Delegitimising Potential of the “Friend-Foe” Axiological Opposition (Exemplified by English-Language Political Internet Comments)

The article considers the axiological opposition “friend-foe” on the material of non-professional political discourse. English-language comments to the posts of Bill de Blasio, Mayor of New York City, and Sadiq Aman Khan, Mayor of London, published in 2016–2021 in social networks are analyzed in the linguapersonological aspect. The speech techniques of expressing the “friend-foe” binary used in discrediting the current government by users of social networks are revealed. The authors of the comments use various means of verbal aggression to undermine the legitimacy of the authorities. At the lexical level, invective and pejorative units are used. At the grammatical level, the users have recourses to imperative constructions and the superlative degree of adjectives with negative evaluation. Internet users’ comments to politicians’ posts in social networks are shown to form the image of a political actor on the basis of the dominants ‘deviant behaviour’, ‘bad manager’, ‘foe, not their own’, which are determinants in discrediting the authorities. Delegitimising characteristics-determinants used in American and British non-professional political discourse are revealed to be similar due to the general unification and globalization trends. When writing a negative response, the universal institutional characteristics of a politician come to the fore. The unique individual and personal characteristics of mayors are also found to be important in creating a more voluminous image of an illegitimate leader.

Genres of New Media in Linguistic Expertise Practices

The article addresses the debatable points of research into present-day media genres: memes, surveys, comments, social network posts, etc. that become the materials for forensic linguistic expertise. The issue of identifying the object of linguistic expertise of information materials is discussed. The ways the new discursive reality transforms the existing methodological approaches are conceptualized. The brevity of the texts under examination, the dependence on the sociocultural context and the communicative situation, the presence of hypertext determine the complexity of the expert assessment of new media materials. The cases of ambiguous interpretation of brief comments and the peculiarities of defining the subject of speech in them are dealt with in the article. Special attention is paid to the techniques of play on words (pun) and irony as well as the linguistic semiotic analysis of creolized texts. The problem areas of expert research into new media materials included in the space of social communications are characterized. Special attention is paid to the problem of multi-discipline expertise, since the specificity of the material under study and the tasks to be solved determine the involvement of specialists of the socio-humanistic profile (psychologists, sociologists, political scientists, religious scholars, etc.) and necessitate solving the issue of the boundaries of experts’ competence. The new discursive reality associated with the involvement of mass audience in the production of media content suggests some changes in approaches to the linguistic expert assessment of the materials under scrutiny.

How Credible Is the Central Bank of Russia to Russian Households?

The article examines the degree of public trust in the Central Bank of Russia after the transition to inflation targeting. For this purpose, the author analyzes available data from inFOM consumer surveys of confidence in the Bank of Russia as a whole as well as in its inflation targets from 2014 to 2024. Public confidence in the Bank of Russia remained relatively high from 2014 to 2017 as inflation targeting was adopted, but the Bank became far less credible to households from 2019 to 2022 because they did not believe the inflation target of 4% was achievable by the end of the current year or over the medium term. This is probably due to their lack of awareness that the Bank’s principal goal is price stability even though they are actually experiencing high inflation. The households surveyed think that government authorities are largely responsible for controlling inflation. Furthermore, consumers do not take notice of the information provided by the Central Bank of Russia. Using news articles from Russian media, posts from news communities on the VKontakte social network, and Google Trends search queries, the author shows that consumers who ignore information about the Bank’s policy are at a serious disadvantage when there is economic uncertainty. The discrepancy between the inflation expectations of consumers and professional forecasters diminishes as the volume of information tied to the phrase “Bank of Russia” increases in media, social network posts, and online searches.

Disinformation and "hate speech" in the online media of Kazakhstan (results of monitoring 2022)

The article examines the problem of using the "language of hostility" and disinformation in the media texts of the Kazakh media, as well as the problem of stuffing fake news during the political events of January 2022 in Kazakhstan. The goals of disinformation are determined and the effects on the mass audience are evaluated. During the tragic events at the beginning of the year, the Internet was turned off in the country, which led to an information vacuum. After the restoration of communication, a stream of fake materials of various formats (audio, video, images and messages on social networks) began, the purpose of which was to incite fears among the population. The platforms WhatsApp, YouTube, Telegram and VKontakte were investigated. Case analysis of texts was also used as part of the analysis. The materials were selected by random sampling. Various criteria were taken into account for the selection of publications, including attendance, the toxicity of comments, the number of articles and posts in online publications and social networks, as well as genres of materials and the presence of elements of hate speech. The purpose of this study is to comprehensively analyze the destructive consequences of disinformation recorded in the Kazakh and Russian media during the attempt to forcibly seize power in the Republic of Kazakhstan. The following methods were used: critical analysis of scientific literature, media monitoring, discourse analysis, and the method of comparative analysis. Case analysis of texts was also used. In 2022, Kazakhstan faced new geopolitical and information challenges: armed border conflicts, disinformation and "hate speech". All these risks have led to an even greater polarization of Kazakh society, socio-political tension and a decrease in the level of trust in the official media. The key factor in changing the situation may be increasing the information literacy of the population and its promotion through cooperation with civil society institutions, the implementation of universal media education programs. Given the volume of data and the amount of information processed, it is difficult to fit the whole situation and analyze how the discourse of "hate speech" and disinformation was built. The repeatedly disseminated unreliable and contradictory information significantly increased the tension of the network media field, contributing to a greater polarization of opinions about the extraordinary event and the aggression of participants in the information exchange.

ABOUT BUSINESS E-MAIL AS A GENRE OF PROFESSIONAL INTERNET COMMUNICATION IN THE COURSE OF “LANGUAGE OF BUSINESS COMMUNICATION”

The research focuses on exploring lexical-grammatical, syntactic, punctuation features of the text of business e-mail as a genre of professional Internet communication in the course of “Language of business communication”. The tasks are outlined as follows: 1) to reveal the contents of the course of “Language of business communication”; 2) to highlight the importance of the content module on professional Internet communication; 3) to analyze business e-mail as a genre of professional Internet communication. The descriptive method and various types of linguistic analysis are applied. Emphasis is placed on the motivational-practical (speech) component of the course as a defining feature of the “Language of business communication” discipline, which determines the approach to the thematic content of the four content modules. The proposed research highlights the fourth content module, which reveals the peculiarities of Internet business communication and encompass follows themes: “Internet communication: specifics, types, genres”, “Language organization of business Internet communication”. The analyzed content module involves familiarizing students with the concept of Internet style, types of network communication, genres of business Internet communication, types of posts in social networks, and the culture of Internet communication. To improve the practical skills of business electronic correspondence, an educational model (formulated in linguistics) was used, which consists in the synthesis of four interrelated aspects. The first aspect is based on mastering the theoretical foundations of the researched problem: philology students reveal the essence of the concept of an electronic business letter, systematize the basic principles of creating and rules for writing business e-mails. The second aspect is practical. It consists primarily in the formation of skills and abilities to analyze the texts of business letters. For example, the structural-compositional, lexical-grammatical, and punctuation features of an electronic Cover Letter and a Request Letter are considered. The third aspect is research. It involves independent processing of the text of an electronic business letter in accordance with the proposed algorithm. The fourth stage is creative. It provides for the independent creation of an electronic business letter taking into account the specified requirements. The results of the research can be used as materials for practical classes on the language of business communication, Ukrainian language for professional purposes, language culture, etc. Key words: “Language of business communication”, business e-mail, professional Internet communication, etiquette of business correspondence.

Digitalization of cultural industries: Evidence from the official Spider-Man movie TikTok account

The digitalization of culture and creative industries has presented both challenges and opportunities. To effectively engage audiences and ensure their constant interest, these industries must continually adapt their circulation strategies. The objectives of this study are threefold: 1) analyze the strategies employed by the Marvel Cinematic Universe (MCU) to engage with the fan community; 2) examine the interplay between fiction and reality based on the relationships between characters and actors, as shown in social network posts; and 3) determine how the discourse in the film ‘Spider-Man: No Way Home’ is reconstructed on TikTok, leveraging its particular ‘affordances’ (Gibson, 1979). A mixed methodology approach was utilized, combining the use of Analisa.io software to provide a comprehensive view of the Spider-Man official account’s activities and user engagement with a detailed analysis of the thematic, multimodal, and discursive aspects of TikTok videos published by the account. The findings reveal that the MCU employs TikTok as a marketing tool for the new movie, utilizing three main strategies that align with the characteristics of transmedia storytelling and enhanced audience engagement: 1) strategic selection of publication moments for the growth of follower engagement; 2) extensive use of teasers; and 3) leveraging the presence of actors as a key element to facilitate fan identification.

Discovering why people believe disinformation about healthcare.

Disinformation-false information intended to cause harm or for profit-is pervasive. While disinformation exists in several domains, one area with great potential for personal harm from disinformation is healthcare. The amount of disinformation about health issues on social media has grown dramatically over the past several years, particularly in response to the COVID-19 pandemic. The study described in this paper sought to determine the characteristics of multimedia social network posts that lead them to believe and potentially act on healthcare disinformation. The study was conducted in a neuroscience laboratory in early 2022. Twenty-six study participants each viewed a series of 20 either honest or dishonest social media posts, dealing with various aspects of healthcare. They were asked to determine if the posts were true or false and then to provide the reasoning behind their choices. Participant gaze was captured through eye tracking technology and investigated through "area of interest" analysis. This approach has the potential to discover the elements of disinformation that help convince the viewer a given post is true. Participants detected the true nature of the posts they were exposed to 69% of the time. Overall, the source of the post, whether its claims seemed reasonable, and the look and feel of the post were the most important reasons they cited for determining whether it was true or false. Based on the eye tracking data collected, the factors most associated with successfully detecting disinformation were the total number of fixations on key words and the total number of revisits to source information. The findings suggest the outlines of generalizations about why people believe online disinformation, suggesting a basis for the development of mid-range theory.

Linguistic Models of Mass Media Genres` Stylistic Diffusion in the Communicative Space of Ukraine and Bulgaria

Research attention is focused on fundamental changes in techniques and methods of informing , which activate the communicative capabilities of language and give rise to new genres, forms and formats of interpersonal communication in various spheres and discourses. It actualizes the research problem of the diffuse stylistic nature of mass media genres, argued by the processes of assimilation of public communication`s spheres. The author’s argumentation is based on the vectors of the dynamics of the literary language`s evolutionary potential, that is focused on the format and tasks of social communication. Predictably, the architectonic heterogeneity and polyfunctionality of the language of media texts explain the syncretic nature of the genres. And also the tendency to their constant modification and transformation regarding the structure, meanings and communicative valence of language units. The article interprets models of updating of the functional and stylistic paradigm of modern literary languages in their genre projection and perspective, which do not depend on the time and space of culture. To prove this thesis, the texts of the Ukrainian and Bulgarian media space in the genre of posts in social networks were chosen as the object of analysis. The main research aim is to determine the typicality / atypicality of stylistic diffusion in the genre paradigm of Ukrainian and Bulgarian media. The attention is paid to the bloggers and companies that are well- known in Ukraine and Bulgaria as influencers who make professional activities in the fields of health, beauty, sport, etc., and really model the current mass language consciousness. The research base is formed by texts that were published in the social networks Instagram, Facebook in 2023 – 2024. The functional and stylistic methodology of the mentioned material`s interpretation allows in a comparative discourse to prove the typicality of communicative processes that are resulted in the stylistic diffusion of modern mass media genres.

Marijuana Use and Perceptions of Employment Suitability

Abstract: The present study extends research on marijuana social acceptance and employment suitability in the United States by examining how hiring managers view substance use-related content in social networking profiles. Specifically, this research focused on the weight hiring managers placed on social networking posts containing recreational marijuana content, medicinal marijuana content, and alcohol content when assessing potential candidates. With a sample of 405 hiring managers who evaluated experimentally manipulated social networking profiles, the results demonstrated a modest negative hiring bias against recreational marijuana content, with stronger negative effects for alcohol content. At the same time, social networking content related to medicinal marijuana content did not have a significant effect on perceptions of employment suitability. These findings highlight the nuanced nature of substance use stigma in today’s evolving society.

Tests of historical, cultural and patriotic orientation for student public posts in social networks: the experience of creating

The article contains examples of creating tests for students timed to memorable dates and posted in the public of structural divisions of universities in social networks as a tool of historical and cultural education and the formation of patriotism.

TEACHING SIMULTANEOUS TRANSLATION IN SOCIAL NETWORKS TO FUTURE INTERPRETERS

The article is devoted to teaching the main features of simultaneous translation in social networks, taking into account the specifics of this genre of communication. The relevance of the issue is that now more and more people are using social networks as a means to promote their business, and thus many experts and brands need translators, and especially simultaneous ones, who could help adequately represent products and services on the international market. The purpose of the article is to analyze the main challenges that translators face when working with text messages, comments and posts in social networks, namely the speed of reaction, the importance of context and non-standard formats. In addition, the peculiarities of translation in various social networks, such as Twitter, Facebook, Instagram, LinkedIn, and their influence on the process of simultaneous translation are considered. In the course of the research, such methods as observation, synthesis and analysis of information were carried out. The article presents an analysis of the main features of simultaneous translation, namely speed, brevity, informality, a large amount of visual content, emotional aspect, tone of voice and their structural-grammatical and lexical functions. The article contains recommendations for translators on effective and accurate simultaneous translation in social networks. Informational and explanatory functions, to which the whole article is devoted, are only its secondary purpose. The following main methods of simultaneous translation were also highlighted: functional substitution, transcription, transliteration, descriptive translation (explication), reduction, expansion and borrowing. In addition to this, we identified the main approaches used in simultaneous interpreting, namely targeting real-time, blogger, audience, social network and content. In addition, key aspects have found their way into translator training profiles and quality standards for translation services. Central to the acquisition of translation competence is reflection on decisions and actions during the translation process, as well as various process studies.

Positioning analysis of Spanish politicians through their Twitter posts versus Spanish public opinion

The evolution of Information and Communication Technologies (ICT) has changed the way we communicate. Access to the Internet and social networks has even changed the way we organise ourselves socially. Despite advances in this field, research on the use of social networks in political discourse and citizens’ perceptions of public policy remains scarce. So, the empirical study of politicians’ discourse on social networks in relation to citizens’ perception of public and fiscal policies according to their political affinity is of particular interest. The aim of the research is, therefore, to analyse positioning, from a dual perspective. Firstly, the study analyses the positioning in the discourse of the communication campaigns posted on social networks of Spain’s most prominent politicians. And secondly, it evaluates whether this positioning is reflected in citizens’ opinions about the public and fiscal policies being implemented in Spain. To this end, a qualitative semantic analysis and a positioning map is performed on a total of 1553 tweets published between 1 June and 31 July 2021 by the leaders of the top ten Spanish political parties. In parallel, a cross-sectional quantitative analysis is carried out, also through positioning analysis, based on the database of the Public Opinion and Fiscal Policy Survey of July 2021 by the Sociological Research Centre (CIS), whose sample is 2849 Spanish citizens. The results show a significant difference in the discourse of political leaders’ social network posts—which is more pronounced between right-wing and left-wing parties—and only some differences in citizens’ perception of public policies according to their political affinity. This work contributes to identifying the differentiation and positioning of the main parties and helps to guide the discourse of their posts.

Understanding What Drives Long-term Engagement in Digital Mental Health Interventions: Secondary Causal Analysis of the Relationship Between Social Networking and Therapy Engagement.

Low engagement rates with digital mental health interventions are a major challenge in the field. Multicomponent digital interventions aim to improve engagement by adding components such as social networks. Although social networks may be engaging, they may not be sufficient to improve clinical outcomes or lead users to engage with key therapeutic components. Therefore, we need to understand what components drive engagement with digital mental health interventions overall and what drives engagement with key therapeutic components. Horyzons was an 18-month digital mental health intervention for young people recovering from first-episode psychosis, incorporating therapeutic content and a private social network. However, it is unclear whether use of the social network leads to subsequent use of therapeutic content or vice versa. This study aimed to determine the causal relationship between the social networking and therapeutic components of Horyzons. Participants comprised 82 young people (16-27 years) recovering from first-episode psychosis. Multiple convergent cross mapping was used to test causality, as a secondary analysis of the Horyzons intervention. Multiple convergent cross mapping tested the direction of the relationship between each pair of social and therapeutic system usage variables on Horyzons, using longitudinal usage data. Results indicated that the social networking aspects of Horyzons were most engaging. Posting on the social network drove engagement with all therapeutic components (r=0.06-0.36). Reacting to social network posts drove engagement with all therapeutic components (r=0.39-0.65). Commenting on social network posts drove engagement with most therapeutic components (r=0.11-0.18). Liking social network posts drove engagement with most therapeutic components (r=0.09-0.17). However, starting a therapy pathway led to commenting on social network posts (r=0.05) and liking social network posts (r=0.06), and completing a therapy action led to commenting on social network posts (r=0.14) and liking social network posts (r=0.15). The online social network was a key driver of long-term engagement with the Horyzons intervention and fostered engagement with key therapeutic components and ingredients of the intervention. Online social networks can be further leveraged to engage young people with therapeutic content to ensure treatment effects are maintained and to create virtuous cycles between all intervention components to maintain engagement. Australian New Zealand Clinical Trials Registry: ACTRN12614000009617; https://www.australianclinicaltrials.gov.au/anzctr/trial/ACTRN12614000009617.

Jair Bolsonaro and the defining attributes of the populist radical right in Brazil

Abstract The presidential candidacy of Jair Bolsonaro in Brazil in 2018 was the Populist Radical Right’s (PRR) first one to be electorally successful in Latin America. Although it is possible to assert that Bolsonaro belongs to the PRR given his political career, few studies have focused on the supply-side and Bolsonaro’s ideas and discourses. This article analyzes the relevance of the three defining attributes of the PRR – nativism, authoritarianism, and populism – in Bolsonaro’s ideological and programmatic platform. Based on an analysis of the 2018 electoral manifesto, campaign and government speeches, and social network posts, three arguments will be made: (i) authoritarianism is the primary defining attribute of Bolsonaro; (ii) populism only exists in the country associated with a solid negative political identity, an anti-Workers’ Party sentiment known in Brazil as antipetismo; and (iii) neoliberalism must be considered a fourth ideological element of the Brazilian PRR.

IMPLEMENTATION OF THE INTELLECTUAL SYSTEM OF SENTIMENT ANALYSIS AND CLUSTERIZATION OF PUBLICATIONS IN THE TWITTER SOCIAL NETWORK

Thanks to the intensive development of social networks, the intensity of exchange of short electronic text messages is constantly increasing, the tone of which can serve as a sensitive indicator of public mood and important social phenomena, interesting for sociologists, politicians, economists, and specialists in other fields. In this regard, the task of automating the processing of such natural language messages is of significant scientific and practical interest. The object of this study is the sentiment of user publications in the Twitter social network. Due to the great popularity of the social network itself and the large number of user messages, which are short in nature, it is possible to conveniently determine the mood of user posts and combine them into clusters according to the given parameters of the intelligent system. The subject of the study is methods and algorithms for analysing the sentiment of large arrays of messages containing the necessary keywords and relating to a certain specific topic, determining the factors and distributions of the sentiment of messages based on the input array of system data, dividing messages into main groups and providing estimates within certain defined limits in to each group, division into clusters according to the obtained search point and display of the obtained results in the desired format. The purpose of the work is to implement an intelligent system of sentiment analysis and clustering of publications based on a recurrent neural network of long short-term memory (LSTM) and the k-means clustering algorithm. The following main tasks are solved in the work: 1. To analyse the most used and newest algorithms, methods, approaches and means of implementing tasks of sentiment analysis and clustering of publications in social networks. 2. To develop a conceptual structure of an intellectual system of sentiment analysis and clustering of publications. 3. To form functional tasks for the key modules of the created intelligent system of sentiment analysis and clustering of publications in the Twitter social network. 4. Implement an intelligent system of sentiment analysis and clustering of publications based on a recurrent neural network and the k-means clustering algorithm and conduct experimental verification. Among the methods used for this purpose are the recurrent neural network of long short-term memory; k-means clustering algorithm. The following results were obtained: the general structure of the intellectual system of sentiment analysis and clustering of publications was analyzed, designed and built. The main task of creating the system, first of all, was to improve the recurrent neural network of long-short-term memory, which, thanks to the improved algorithm, significantly facilitates text processing by natural language processors according to text data of a certain size. Also, a special clustering algorithm, namely k-means, was used in parallel, thanks to which it was possible to change the general approach to clustering and the creation of final clusters, in accordance with the obtained results of the work of the recurrent neural network. Conclusions: As a result of applying a combination of LSTM neural network and k-means clustering algorithm, it was possible to speed up the process of sentiment analysis and clustering of posts in the Twitter social network by 10...15% compared to similar convolutional neural networks and hierarchical clustering.

Social media popularity prediction with multimodal hierarchical fusion model

With the development of social media, data are becoming increasingly diversified, including images, texts and videos, which also makes it possible to use multimodal information to predict popularity. To effectively predict the popularity of specific posts in social networks, we propose a social media popularity prediction framework based on a hierarchical fusion model. We first perform feature extraction on four kinds of information for two modalities (image, image attribute, text and text attribute). Here, we extract image features with Residual Nets (ResNet) and generate image attribute features with an image attribute predictor. We use the image attribute features, Global Vectors for Word Representation (GloVe) embeddings and the bi-directional Long Short-Term Memory (LSTM) model to obtain text features, and the text attribute features are added through data preprocessing. Then, the features of the kinds of four information are reconstructed with our multimodal hierarchical fusion model and merged into a feature vector for popularity prediction. Experimental results show that our hierarchical fusion method has excellent performance for popularity prediction on two real-world datasets.

Sentiments Analysis of Amazon Reviews Dataset By using Machine Learning

Abstract: Any opinion of a person that can convey emotions, attitudes, or opinions is known as a sentiment. The data analyzes that are collected from media reports, consumer ratings, social network posts, or microblogging sites are classified as opinion mining research. Analysis of sentiment should be viewed as a way of evaluating people for particular incidents, labels, goods, or businesses. The amount of views exchanged by people in micro-logging sites often increases, which makes nostalgic interpretations more and more common today. All sentiments may be categorized as optimistic, negative, or neutral under three groups. The characteristics are derived from the document term matrix using a bi-gram modeling technique. The sentiments are categorized among positive and negative sentiments. In this analysis, the Python language is used to apply the classification algo for the data obtained. The detailed accomplishment of LinSVC demonstrates greater precision than other algos.

ClarTM: a method for geolocations clarification within extensive urban sites using topic modelling

Determining the correct geolocation of user posts in social networks plays a vital role in many studies investigating and improving the urban environment. We started our research on the possibilities of refining geolocations in extended urban objects, where posts refer to locations exceptionally unevenly, and we continue it in this work, supplementing the location redefinition algorithms with the new methods. Namely, we have developed a three-stage location redefinition algorithm. In the first stage, it expands geolocations from external sources, enriches them with descriptions and combines geolocations from different sources. The second stage redefines posts to locations based on recognising named entities and semantic similarity metrics. And in the third stage, it eliminates inaccuracies in popular geolocations using topic modelling and improves the quality of geolocation clarification by redefining posts to more appropriate geolocations. Topic modelling, which found topics to redefine in 7 (11% of all) locations, increased the redefinition's accuracy by 30% (from 57% to 87%).