Методи і засоби виявлення інформаційно-психологічної маніпуляції в онлайн-спільнотах

Abstract

У дисертаційній роботі розв’язано важливе наукове завдання – розроблення методів і засобів виявлення інформаційно-психологічної маніпуляції (ІПМ) в онлайн-спільнотах, що дало змогу вчасно виявляти прецеденти ІПМ, запобігати їх розповсюдженню, підвищило якість комунікації в онлайн-спільнотах та кількість учасників онлайн-спільноти. У роботі запропоновано та обґрунтовано методи виявлення ІПМ в онлайн-спільнотах та заходи нейтралізації ІПМ з метою уникнення їх згубного впливу та для забезпечення ефективної та успішної комунікації в онлайн-спільноті. Розроблено програмно-алгоритмічний комплекс «Детектор інформаційно-психологічної маніпуляції» на основі формальних моделей онлайн-спільноти та тактики ІПМ, що дало змогу виявляти і нейтралізувати ІПМ та підвищити ефективність та якість загального процесу комунікації в онлайн-спільноті та виявляти ІПМ з точністю 30–40 %. В диссертационной работе решена важная научная задача разработки методов и средств обнаружения информационно-психологических манипуляций (ИПМ) в онлайн-сообществах, что позволило своевременно выявлять прецеденты ИПМ, предотвращать их распространение, повысило качество коммуникации в онлайн-сообществах и количество участников онлайн-сообщества. В работе предложены и обоснованы методы выявления ИПМ в онлайн-сообществах и меры нейтрализации ИПМ во избежание их пагубного влияния и для обеспечения эффективной и успешной коммуникации в онлайн-сообществе. Разработан программно-алгоритмический комплекс «Детектор информационно-психологической манипуляции» на основе формальных моделей онлайн-сообщества и тактики ИПМ, что позволило выявлять и нейтрализовать ИПМ и повысить эффективность общего процесса коммуникации в онлайн-сообществе, выявлять с точностью 30–40 %. In the thesis an important scientific task of the development of the methods and means of detection of information and psychological manipulation (IPM) in online communities is solved. This allowed to identify the IPM precedents in a timely manner, to prevent dissemination of IPM, to increase the quality of communication in online communities and the number of participants in an online community. The thesis proposes methods for detecting IPM in online communities and steps required to neutralize such IPM and as a result, avoid the impact of IPM which is harmful for effective and successful communication in an online community. The first chapter contains a description of the analysis of the online community as a platform for IPM, in particular, the typical structure of an online community, types of the text communication and their characteristic features. Peculiarities of using verbal, nonverbal and paraverbal means in online communities are addressed. Approaches to the classification of online communities are overviewed, on the basis of the latter the research area of the thesis is outlined, namely public online communities with interactive text communication. Comparative analysis of the traditional manipulation and IPM is carried out. Grounding on the found common features, possibilities of adapting existing schemes of manipulation tactics to online communities are discussed. Sentiment analysis tools, which are further applied to identify psychological states of discussion participants, are reviewed. Existing classifications of emotions and psychological states of the human are overviewed. The possibility of utilizing dialog acts tags for the construction of IPM markers is considered. The second chapter is devoted to the analysis of an online community as a platform for IPM. The improved formal model of an online community is suggested. The latter is designed to contain the necessary for detecting IPM data. The formal model of the IPM tactic on the basis of piece-wise linear aggregate is developed. The formal model of the IPM tactics contains IPM markers as well as the formal description of psychological states of community members, which serve as an evidence of applied IPM. Criteria for detecting suspicious IPM fragments are determined. The criteria constitute the basis of filters for detecting suspicious fragments of the discussion. The notion of a semantic variable is introduced and the structure of semantic variables classes is described. The third chapter is dedicated to the development of the algorithm for detecting IPM in online communities. The algorithm consists of four stages: the preliminary stage, the stage of detection, the stage of neutralization, the stage of the formation of results and recommendations. At the first stage of the algorithm, methods and means for finding relevant discussions in the social media of the Internet are used. At the second stage, methods and means for detecting suspicious fragments of the discussion, methods and means for detecting the sequences of psychological states, methods and means for detecting IPM tactics in discussions, as well as instruments of sentiment analysis, are used. At the third stage, methods and means for determining the author of the text, in particular, linguistic, behavioral, and profile characteristics are used. In addition, methods based on the social graph of an online community are applied to identify influential graph nodes as well as the way information disseminates in the community. In the fourth chapter, the software and algorithmic complex for detecting IPM is constructed on the basis of the developed formal models of the online community and IPM tactics. The complex is designed to detect IPM in the particular online community or IPM directed at an organization in the user-generated content of various online communities. The software and algorithmic complex involves performing tasks of detecting IPM with varying degrees of detailing of the initial conditions. This allows to increase effectiveness of detection and neutralization of IPM by 30–40% and increase the quality of the overall communication process in the online community.