Registration Number
0223U005421,
(0123U104030)
, R & D reports
Title
Determining the trustworthiness of public information on social media using linguistic-based text analytics approach
popup.stage_title
Створення анотованого багатомовного текстового корпусу достовірною урядовою інформацією (TIC) та невизначеною інформацією Інтернет контенту з визначеної ПО.
Head
Khairova Nina F., Доктор технічних наук
Registration Date
22-12-2023
Organization
National Technical University "Kharkiv Polytechnic Institute"
popup.description1
In the project, we aim to develop an actionable approach to determine the trustworthiness of public information on social media that harnesses the power of computational and linguistic-based analytics techniques (i) to expand understanding of the degree and nature of public information trustworthiness in social media platforms, and (ii) to determine the core topics characterizing government digital platform or on Twitter government official's account content and information needs of citizens in communication with government.
An incremental practical contribution of the project is that its results will be equally useful both for (i) government representatives to formulate data-driven recommendations and guidelines based on the provided deep text analytics, and (ii) for data scientists and analysts to evaluate and adapt the proposed methodology for analyzing the nature and trustworthiness of textual messages in social media. Additionally, our study allows increasing possibilities to utilize the OdEE method in socially significant domains due to the establishment of dictionaries and patterns of events. Further, a great benefit of the project is processing exactly Polish and Ukrainian languages and the development of the open linguistic resources for these languages. Such linguistic resources as specialized dictionaries and events patterns of socially fascinating domains can be used in the follow-up studies.
Empirically, we plan to apply our approach to guide the interpretation of case study findings by using the following data from Twitter: (1) tweets from official accounts of Polish and Ukrainian ministries and responsible ministers or secretaries; and (2) Polish- and Ukrainian-language tweets related to government policymaking and their implementation in six governmental sectors (education, health, finance, social welfare, labour information and the environment) from non-governmental accounts.
popup.description2
The object of research: methods and models of supervised and unsupervised machine learning used for classifying reliable and unreliable textual information. The project aims to develop a practical approach to determine the truthfulness of public information in social networks using computational and linguistic analytics methods, for (1) expanding the understanding and nature of the credibility of public information on social media platforms, and (2) identifying the main topics of citizens' information needs in communicating with the government on state digital platforms and social media.
Obtained results: (i) the multilingual corpus of reliable and undetermined reliability information has been created and annotated; the corpus includes texts in Ukrainian, Polish, and English languages extracted from social media, news websites, and official government Internet sources on selected topics; (ii) an annotated dataset has been developed based on the corpus; (iii) linguistic pre-processing analysis of texts, including traditional stages of natural language processing pipeline such as sentence segmentation, tokenization, word lemmatization, POS tagging, and syntactic dependency parsing, as well as additional semantic processing of the corpus: Latent Dirichlet Allocation, Topic Modeling, and Sentiment Analysis.
Key constructive and techno-economic indicators, effectiveness: improvement in the quality of semantic analysis regarding reliable and unreliable textual information in Ukrainian, Polish, and English languages, creation of a new annotated dataset for evaluating models classifying true and false information.
Areas of application: security and cybersecurity, government communication with society through social networks, development of digital government applications, creation of programs and projects for media literacy, education, media, and journalism.
Product Description
popup.authors
Babkova Nadiia V.
Lukashevskyi Arsenii M.
Redozub Ivan A.
Sharonova Nataliia V.
popup.nrat_date
2023-12-22
