Decision-Guidance Management Systems (DGMS): Seamless Integration of Data Acquisition, Learning, Prediction and Optimization

Abstract

A decision guidance management system (DGMS) is a productivity platform for fast development of applications that require a closed-loop data acquisition, learning, prediction, and decision optimization. This paper introduces the DGMS concept, and the first DGMS data model with its query language, DG-SQL. The DGMS data model is an extension of the relational model with probability distributions over a set of attributes as random variables. DG-SQL supports a seamless integration of (1) querying the data collection and construction of learning sets, (2) learning from the learning sets, using parameterized transformers and optionally defining an estimation utility, such as sum of squares of errors, to be minimized; (3) probabilistic prediction and simulation, using expressions that involve random variables, such as expectation, variance and probability of a logical formula; and (4) stochastic or deterministic optimization, where search space is defined as a set of feasible non-deterministic query evaluations.