Context-Redefined Language Application for the Tasks of Intelligent Learning Agents (Resource Consumption Behavior Prediction Tasks)

摘要

This work deals with design and application questions of context-redefined computer languages for new information technologies. Realization problems of such languages are discussed for intelligent learning agents, which applied for solving of resource consumption behavior prediction tasks in communal services. The same approach can be used for another task solution too. Intelligent agent has complicated actions, which is defined by environment interaction and by internal states forming. As a result there are problems of intelligent agents design: search and formation of learning algorithms for intelligent agent; variation and invasion of these algorithms into the own intelligent agent; the paper focuses on just the second problem. The approach is in the application of context-redefined language and it support system for problem solution. The main part of the intelligent learning agent is performance element. There are six components of agent performance element [1]. All of them are discussed from the context-redefined language use point of view. We interested in the conditions and methods context forming for every component of performance element. Main idea is to extract changing parts of component algorithms and organize proper interaction between every part and the context which can change it directly or indirectly. As a result, required adaptive algorithm variation takes place on the base of obtained knowledge. The application of this problem is development of analytical software in a complex hardware-software project for calculating energy and water consumption in a communal industry. Analytical software in this project is a separate module and we can use methods of software development which differ from client and server modules.