The paper describes underlying ideas and operation of the natural language processor, which is a part of the Dialog Information Logical System (DILOS; Computer Center of the USSR Academy of Sciences, Moscow). 

The natural language comprehension is assumed to be performed within: (a) general context, determined by the whole system purpose; (b) some local context, connected with current data base. The basic parts of the processor are described: the main program, transition network and current vocabulary as well as special mechanisms provided for tackling homonymy and words, unknown to the system. Some suggestions are proposed for combining the existing system with a system with a syntactical analyzer.