Elements of nonlinear analysis of information streams

• Дата: 2017
• Автори: Lande, D. V., Hraivoronska, A. M.
• Формат: Стаття
• Мова: rus
• Опубліковано: Інститут проблем реєстрації інформації НАН України 2017
• Теми: information streams; time series; nonlinear dynamics; correlation analysis; fractal analysis; forecasting; информационные потоки; временные ряды; нелиней­ная динамика; корреляционный анализ; фрактальный анализ; прогнози­рование; інформаційні потоки; часові ряди; нелінійна динаміка; кореляційний аналіз; фрактальний аналіз; прогнозування
• Онлайн доступ: http://drsp.ipri.kiev.ua/article/view/126538
• Назва журналу: Data Recording, Storage & Processing

The methods of nonlinear dynamics to apply for analysis of time series corresponding to information streams on the Internet are considered. The information stream consists of documents published on the Internet during some time and related to a certain topic. If one gathers such documents with time stamps, then it is possible to define the time series as the amounts of documents published in short periods and analyze how these amounts vary over time.In the main, the methods discussed are based on correlation, fractal, multifractal, wavelet, and Fourier analysis. The article is dedicated to a detailed description of these approaches and interconnections among them. For instance, correlation is a concept of particular importance and a basis for many techniques. On the other hand, information processes are often self-similar; therefore, fractal and multifractal analysis can provide insights into structure and properties of such processes. The methods and corresponding algorithms presented can be used for detecting key points in the dynamic of information processes; identifying periodicity, anomaly, self-similarity, and correlations; forecasting various information processes. The methods discussed can form the basis for detecting information attacks, campaigns, operations, and wars.