This work deals with the design and application questions of context-redefined computer languages for new information technologies. Realization problems of such languages are discussed for intelligent learning agents (ILA), which were applied for solving of resource consumption behavior prediction tasks in communal services. The approach is in the application of context-redefined language and it support system for problem solution. We concentrate attention to principal unpredicted changing of source function algorithms. Built-in context-redefined computer language is an essential tool for this kind algorithms support. We interested in context-redefined language synthesis. This language is used for the conditions and methods context forming for every component of intelligent agent. We pay extra attention to methods of constructive function interpretation, which can be varied or can be also changed. This synthesis is based on the interest to prediction system demands and their variations during functioning. The main idea of built-in language synthesis is to use main parts of the algorithm for ILA components with proper modification by means another algorithms and context connection. Due to this connection, the original algorithm can be changed directly or indirectly in the process of ILA functioning. As a result, We have to extract changing parts of component algorithms and organize proper interaction between every part and the context which can change it directly or indirectly. Required adaptive algorithm variation takes place on the base of obtained knowledge. At the same time, the algorithm must be implemented as quickly as possible, and the language must be simple and clear. Algorithm efficiency is based on flexibility and modifiability of the language.

Keywords: programming languages, embedded languages, context-redefined languages, intelligent agents, computer languages

The submitted paper considers the issues of developing the analytical software which is the part of the hardware-software complex for energy accountability, reporting and analysis, and consumption. The main function of the analytical software is to predict the consumption of energy resources and emergency conditions, as well as non-routine events. To create analytical software, the use of contextually-predefined machine languages is suggested. The issues considered include the implementation of such languages for intelligent agents (IA), which are used to solve problems of resource consumption forecasting in the housing and communal services sector. Particular attention is paid to the methods for supporting the change in implementing functions. The same approach can be used to solve other problems. An intelligent agent features a complex behavior that changes as it interacts with the external environment and is determined by internal states. When creating intelligent agents, there are fundamental problems: - search and formation of algorithms for the training of an intelligent agent, - change and implementation of these algorithms within the agent itself. The work specially dwells on the second problem. The approach is that contextually-predefined languages and implementing systems are used as form-building (morphogenetic) ones. The productive element is the main part of the learning intelligent agent. There are six components of the productive element. They are considered within the context of contextually-predefined languages application. The basic idea is to isolate the parts of the component's algorithm and to arrange an appropriate connection between the particular part and the context that can change it directly or indirectly. As a result, the necessary adaptive change of algorithms is obtained on the basis of accumulated knowledge. One of the most urgent tasks is the form this process during the operation of the IA.

Keywords: software, energy resources accounting, contextually-predefined languages, intelligent agents, machine languages