| Summary: | 碩士 === 國立中正大學 === 會計學研究所 === 93 === In recent years, lots of serious accounting frauds have continually happened all over the world. Top managers in enterprises use various kinds of earnings management skills to embellish the financial statements of the enterprises for various kinds of purposes, whether the enterprises are public or private. Owing to the accounting information of enterprises are quite huge and complicated, the traditional audit technique restricted to the basis of the time, cost, benefit, and manpower, is very difficult to find the behaviors of practicing frauds in enterprises within limited time, even refer to detect in advance. Since the financial statements on the basis of the ongoing-concern convention and kinds of earnings management activities, and characteristic with time serials, we can accomplish the reasonably systemic conjecture via various kinds of external and inherent factors.
This research uses one kind of strong analytic technology－Neural Networks, and proposes an innovative hybrid financial analysis model in predicting enterprise financial crises. Diagnosing the financial statements information in the past three periods offered from enterprises. Simultaneously use trend and static analysis technologies that consider both cross and vertical sections to execute the detailed track analysis. This whole system can learn the risk patterns from the history experience whenever and wherever possible, so as to adapt various kinds of changes immediately, like environment changes, new fraud tricks, law changes, and any others, and then detect enterprise risks more flexibly and efficiently than before.
|
|---|
