隨著網際網路的使用普及化,相對的網路的問題也很來越多,而現今一切講求速度的時代,如何能快速又能自動發現的故障問題,也愈來愈被重視的趨勢。如果網管人員能夠藉著自動診斷異常系統,隨時監控電腦網路的行為,並且能在重大問題發生之前就能做出適當的處理,那麼企業的營運就不會因為電腦網路的問題而造成損失了。 本研究的目的是建置一套利用LOG資料以規則式知識法(Rule-based Knowledge)推論方式的電腦網路異常診斷系統,在企業環境的區域網路裡,把用戶端電腦網路故障的一些原因及狀況,將這些故障資訊收集起來,配合網管人員查看每個節點網路設備的log紀錄及系統Event Log的訊息紀錄,彙整重要之控制要項,建構成一套互動式的專家推論診斷系統。在符合研究預期的成果後,進而達到網管人員能及時的解決問題,提升工作效率,也降低工作負擔。
As more and more people are using the Internet, networking-related problems are also increasing. We can no longer afford using outdated time-consuming ways to repair all these broken hardware. As a result, we need fast and automatic ways to spot errors. If we can develop an automatic diagnose system for the MIS personnel so they can repair potential problems before they make serious harm, we can save great networking-related costs for the business. This research develops a network anomaly diagnosis system using the system log and a rule-based inference mechanism. We collect all user nodes’ anomaly information within the enterprise-wide intranet. This collection of information, combined with the MIS personnel’s repairing logs and system event logs, enable us to create an interactive deductive diagnosis expert system. The end product meets our expectations and also helps MIS personnel to solve questions on time, increase efficiency and reduce their workload.