This page is a placeholder while the SELDON website is designed and prepared. Soon we will replace this with a page that includes links to papers, presentations, reports, plans and other project information for public disclosure.
Automated/Synthetic intelligence generation from massive and diverse social media (SIG)
Applied/automated intelligent psychophysical analysis (AIPA)
(Foundations: PSED Psychosocioeconomic Dynamics)
It is possible to produce accurate information regarding the present and future psychological and physiological states of individuals and groups, including moods, temperaments, dispositions and intentions, through the analysis of digital media created and disseminated across electronic media including multiple types of social networks.
This information is produced by the analysis of features in three types of media: (1) specific media objects (e.g., photos, videos, non-photo images such as graphics and icons, audio messages and texts), (2) sequences of such types of media linked with the same individuals or groups and involving changes in time and location, and (3) comparisons among such media collected over sequences of time and in different locations, from among many different individuals.
This process of knowledge extraction makes use of multiple, parallel forms of machine learning including rule-based and statistically-based algorithms and systems. There are parallel, competing agents (methods) applied to the data streams and the results are integrated by the use of the same fundamental types of learning algorithms, thereby creating a set of probabilistic statements.
The resulting information derived from such massive media analytics is the basis for generating probabilistic inferences regarding the behaviors of both individuals and groups within different physical, psychological, social, and economic contexts. Such contexts are of two types: (1) actual historical situations, and (2) projected response to different types of stimuli and environmental conditions (including those of a socioeconomic nature).
This information is of value to the respective individuals and groups, for their own interest and use. It is also of value to client-subscribers (companies, institutions) wishing to make decisions based upon projected response and behaviors by their target audiences. This value extends through multiple markets and contexts including large-scale financial, political and other social-impact processes.
Note: There is an extensive body of technical papers and other publications and presentations, available upon request. These pertain to work ongoing within areas of synthetic intelligence ("AI" and nmachine learning), datasets and informatics, modeling, and particularly, the developments using VAE, GAN and related algorithms for larege-scale data error-conrrection and relication/simulation. Contact us anbd learn more,.
Martin Joseph Dudziak