Construction of Dynamical Systems for Analysis of Users Flow on Digital Platforms
Abstract
Digital platforms are among the most important economic and social developments of our time. Digital platforms bring together people, organizations and resources for information, services and other exchanges. This project proposes a new -- dynamical systems -- approach to the study of digital platforms. The current game-theoretic approach helps practitioners to optimize their pricing and other short-term strategies. Dynamical systems approach allows the study of some fundamental properties that govern dynamics of platform users and helps platform owners to analyze platforms in several critically important directions discussed below. One of the main factors of a platform s efficiency is the volume of users interacting through the platform. The central subject of our study is the global picture of the dynamics of the volume of platform users. The dynamical phase portrait allows the prediction of future behavior and the tendency of the trajectories of each group of platform users. This dynamics can be viewed from platform owner s point of view. Understanding the properties of the system helps practitioners to optimize the platform efficiency via incentives adjustment. Furthermore, platforms are becoming increasingly complex. This complexity is influenced by the growing number of distinct groups of users (sides of the platform) that interact with each other through the platform. For example, after establishing the trading platform for sellers and buyer, Amazon announced three new product offerings: Elastic Compute Cloud, which lets programmers rent computing capacity on Amazon s systems; Simple Storage Service (S3), which provides cheap access to on-line storage; and Mechanical Turk, which connects firms with people who perform small tasks that are difficult to automate. The multi-dimensional dynamical systems technique is suitable for extension of the study of two-sided platforms, which were more common previously, to multi-sided platforms, which are becoming increasingly more common now. The objective of this proposal is to use platform-related data for the construction of the dynamical system that describes the volume of two or more types of users interacting through the platform. The preliminary study shows that planar differential equations can be constructed with high accuracy from the simulated data. This method of construction is expected to work well with real-life data. Through this project, it will be expanded to the construction of differential equations of dimensions higher than two, which would help to model platforms with a high number of sides. Qualitative analysis of the phase portraits generated by these models will also be performed to understand the long term tendency of users dynamics. Although this project focuses on models for the study of digital platforms, the new method of construction of differential equations from data has applications in other domains, where dynamical systems models can describe physical, chemical, biological and other processes. The long term goal of this work, which can be developed from this proposal, is to extend the dynamical systems approach, initiated in the principal investigator s publications ``Dynamics of two-sided markets and ``UsersÕ dynamics on digital platforms , to the study of digital platforms. Construction of specific dynamical system models with the help of this new proposed technique combined with the ideas from the above publications, will help to explore new properties of these models and will help to create methods for enhancing the efficiency of two-sided and multi-sided platforms. The dynamical systems approach will help to define and optimize generic functions that govern platforms dynamics.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 24, 2019
- Source ID
- W911NF1910399
Entities
People
- Victoria Rayskin
Organizations
- Army Contracting Command
- Tufts University
- United States Army