Abstract: Queue management in websites has not take into account the importance of pages. This paper proposes a Revenue Oriented Queue Management(ROQM) method, determines the trade probability of every page by analyzing the Web server logs, and ranks requests with different priorities. This method ensures that requests with high trade probability will be dropped less, thus leads to more revenue. Experiments show that ROQM will add to 59.2% of trades at most.

Key words: queue management, E-business, revenue oriented, trade probability

摘要： 现有的网站请求队列管理策略没有考虑各页面的重要性。针对电子商务网站，该文提出一种以收入为导向的队列管理策略，通过对日志文件的统计分析，根据各页面最终跳到交易页面的概率，给予请求不同的优先级，优先处理最有可能发生交易的请求，提高交易成功率，保证网站赢得更多的收益。实验结果表明，收入导向的队列管理策略最多能增加59.2%的用户交易量。

关键词: 队列管理, 电子商务, 收入导向, 交易概率

CLC Number: 

• TP393