Intelligent analysis system for monitoring the spread and influence of political internet memes

Abstract

The paper explores the options for developing artificial intelligence technologies and tools aimed at enhancing the functionality of the system designed for monitoring and analyzing political Internet memes in the Russian-speaking segment of the Internet. Achieving this goal on the basis of deep machine learning methods and neural network technologies involves dealing with a number of correlated systemic interdisciplinary tasks (problems), which we discuss in the article. We introduce the architecture and functionality of the Memometrix software package, which can provide a partial solution to the problem of automatic image acquisition, tagging, filtering, grouping and exporting. Th e rationale for distinguishing different levels of macro- and micromodels, models of information interaction between intelligent agents, and control models, has been provided. Simulation modeling is helpful in adjusting the sets of algorithms to be then equipped by a search intelligent agent designed for monitoring and collecting Internet memes, and their analysis, as part of a multi-agent system for controlling the flow of Internet memes. Creating an intelligent system for monitoring Internet memes, its tuning and verification to real processes require the use of tagged databases. Building Internet meme databases involves accumulating and classifying Internet meme images, identifying their duplicates. It is required to use fi ltering, grouping and exporting the data relevant to Internet memes. Searching for memes involves making search queries and an option to perform a reverse image search, in that allowing to use the already existing databases for classifying meme images. We provide a detailed description of the package’s relevant modules, as well as an example of an interface for filtering Internet memes based on using the templates agreed upon with sociologists. The software product was developed with the use of modular architecture and comprises several applications: a Web application (user interface), a server application (provides the operation of the package), and an application for automatically collecting Internet memes. The modularity of the package allows for making adjustments in the application (analysis of collected results, visualization of collected information, prediction of the further spread of Internet memes).