数字电影发行相似案例快速检索方法  

Fast similar case retrieval method for digital film distribution

作　　者：宋双[1] 吴宏涛[2] 高强[1] 刘翼光[2] 

机构地区：[1]北京航空航天大学电子信息工程学院,北京100191 [2]国家广电总局电影数字节目管理中心,北京100088

出　　处：《计算机工程与设计》2011年第2期638-641,共4页Computer Engineering and Design

摘　　要：为解决传统相似案例检索方法在数字电影大型案例库中检索效率低下的问题,针对数字电影大型案例库进行了研究分析,从提高案例检索速度和参考价值的角度出发,提出了一种聚类、优选和匹配相结合的相似案例检索方法。通过聚类缩小案例的检索范围,通过优选在相应的检索范围内形成参考价值较高的候选案例集,通过匹配算法在候选案例集中寻找最相似案例,并通过数字电影发行实施实例表明了其可行性。实验结果表明,该方法提高了案例的检索速度和参考价值。The traditional case retrieval method inefficient to search a case in the large digital film case database, to improve case retrieval speed and the reference value, a three-step case retrieval method is proposed for the large digital film case database. This method includes clustering, optimization and matching. Through the clustering algorithm to narrow the scope of case retrieval, and by optimizing to form candidate of cases with higher reference value, finally in the candidate by matching algorithm focus on finding the most similar case. The implementation of digital film distribution demonstrates its feasibility. Experiments show that this method improves the case retrieval speed and reference value.

关 键 词：案例检索 聚类 神经网络 匹配 数字电影 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]