| 摘要 |
1. Apparatus for monitoring and analyzing the behaviour of a message network comprises a plurality of expert systems (Consultants), each of which is concerned with a different one of the several operations normally involved in the analysis of any given network problem, and a communication means, preferably in the form of a Blackboard 21 by means of which deductions can be passed between the Consultants under control of the user 12 (who may also add information to the Blackboard). In a preferred embodiment, a Fault Recognizer Consultant 13 is fed with raw data from the network and generates Problem suggestions therefrom; a Usage Advisor Consultant 14 suggests where to test and what to test for; an Operations Assistant Consultant 15 suggests how to make the tests; a Expert Commentator Consultant 16 analyses the results of the tests to produce Symptoms; and a Diagnostician 17 generates possible and plausible diagnoses of the observed fault conditions, together with explanations. In practice there in likely to be a roughly cyclic sequence of activation among the Usage Advisor, the Operations Assistant, and the Expert Commentator, with the Diagnostician performing a kind of supervisory or higher level role. <IMAGE> |
