Summary: | 碩士 === 大同大學 === 資訊工程學系(所) === 93 === This paper presents an intelligent inference system for revenue assurance of a telecommunication company, where artificial intelligence (AI) techniques such as grey prediction, fuzzy logic (FL), genetic algorithms (GAs), and neural networks (NNs) will be combined to achieve a more accurate problem detector with a higher availability than those traditional risk management and audit approaches could provide. In the telecommunication companies, revenue leakage can take the form of actual loss like billing errors, fraud and bad debts, or from opportunity loss like billings foregone through incorrect calls. It costs hundred millions of dollars loss annually for each company. However, most practical problems in revenue assurance are complex, full of uncertainty and can only be solved by human resources. Due to such a difficult situation, the problems cannot be effectively prevented and overcome.
A better prediction model from grey system is introduced and different AI techniques for synergism of FL, GAs, and NNs are presented. The optimization mechanisms for our prediction model and the measures of performance with precision and coverage are also discussed. The thesis finally exploits the empirical errors of revenue leakage for measuring prediction ability by using our proposed methods to demonstrate the system potential of revenue assurance.
|