期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2017
卷号：17
期号：9
页码：85-94
出版社：International Journal of Computer Science and Network Security
摘要：Finding the similarity between two objects is the most important fundamental operation in database management as well as in web searching environment. The similarity between two objects is generally computed based on the attribute values of the objects. Traditional similarity measures using only attribute values. In the proposed method, similarity between two objects is the most effective and accurate when the similarity is computed based on the attribute values as well as the voting/rating/preferences of values of attributes. That is, the similarity between objects is not based only on attribute values but instead object similarity is computed based on some of the weighted values of voting/rating/preferences and values of attributes. A linear similarity function is the simplest model for finding weighted similarities between objects. Similarity measure techniques are very much useful in processing database queries such as top-k queries, reverse top-k queries, k-nearest neighbor queries and other different types of queries related to business sales activities. Sometimes there is a need to construct and use multidimensional indexing data structures for efficient search/access of data of very large database sizes and the effective execution of queries. Also sometimes different types of pruning techniques are required for scalability purpose. In general, linear function computations are scalable for similarity finding measurements between objects.
关键词：Similarity between objects; similarity search computations; top-k queries; reverse top-k queries; and other k-nearest neighbor queries; database queries; database operations; attribute values; customer voting/rating/preferences.
