The Study On Data Mining Algorithm And Application In Web Log Analysis

Web log mining is an important aspect of data mining research, aiming at adjusting web site structure, optimizing service performance; responding to every user's special demands; offering individual service and so on. Web log mining is usually compartmentalized three phases, data preprocessing, pattern identifying, pattern analyzing.The paper analyzes the process of data preprocessing in details, data cleaning, user identification, conversation identification, path supplement, and transaction identification, brings forward a method of conversation identification basing on time interval, uses absolute evaluation standard to validate its high effect quality. Against native storage and agent server storage, the paper puts forward new path supplement algorithm, and use test cases to prove its efficiency.The paper lucubrates Fuzzy C-Means(FCM) clustering algorithm in data mining, because of existent disadvantages, suffering influences from isolated points and clustering amount, offers improved Fuzzy C-Means clustering algorithm, add power value for data object's fuzzy subjection degree, and import fuzzy clustering validity function to optimize clustering amount in improved algorithm. The paper carries through instance validating, proving improved FCM to be efficient and possess high performance. Finally, the paper exerts improved FCM to mine and analyze web log, proving its efficiency again.And put forword method using clustering that carries out individuation recommendation web pages sets based on web log mining.