UDC 004.8
DEVELOPMENT OF A MECHANISM FOR COLLECTING AND ANALYZING DATA TO ASSESS THE LEVEL OF TRUST IN GOVERNMENT USING VOICE ASSISTANTS
O. D. Kazakov, Ph.D (in economic sciences), assistant professor, Head of the Department, BSETU, Bryansk, Russia;
orcid.org/ 0000-0001-9665-8138, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. K. Sologubov, master student, BSETU, Bryansk, Russia;
orcid.org/0000-0001-7878-5838, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The technologies of voice assistants for public opinion research are considered. The aim of the work is to implement a mechanism for assessing the level of trust to the government using the results of voice assistants dialogue analysis based on natural language processing methods and simple multilayer fully connected feed forward neural networks. To complete the task of forming labeled training and test data set necessary for training a model to assess public opinion and the level of trust in government authorities, the skill of a virtual assistant from Yandex – Alice was developed. Also, Yandex.Alice's «Political Consultant» skill has been implemented for systematization and critical analysis of data on the level of public confidence in government bodies. To solve natural language processing problems, including the analysis of the sentiment of dialogues, ELMo model from DeepPavlov library, trained on the basis of RuSentiment, has been used. The output of the results to assess the sentiment of dialogues and the level of trust is presented in the form of analytical subsystem implemented in the form of Python scripts and Microsoft Power Bi report.
Key words: voice assistant, assessment of the level of trust in authority, natural language processing, feed forward neural networks, text sentiment analysis.