Index Structure Optimization And Updating Algorithm Based On Growing Network

With the rapid development of modern Internet and digital multimedia technology, a large number of social network sites and instant communication tools have become more and more popular. The multimedia data such as pictures and video cuts are exploding. The data are generally characterized by high dimension and mass. At present, research of the data processing technology is a hot topic in the field of cloud computing and big data. It is very important to study how to manage and quickly query the data in the database which has the characteristics of high dimension and mass. It is one of the effective solutions to study query technology in the high-dimensional indexing. When dealing with huge amounts of high-dimensional data, Traditional high-dimensional indexing technology will show a series of problems:tree structure indexing will have the curse of dimensionality; dimension reduction indexing will lose useful information; hash based indexing is hard to design excellent hash functions. The novel high-dimensional indexing technology based on small-world model can solve these problems effectively and be applied to high-dimensional data. It has a very broad research and development prospect. In this dissertation, the growing small-world network model is applied to the research and optimization of high-dimensional indexing technology. It mainly includes the following aspects:(1) Optimization of the index structure and generation algorithm based on growing small-world network. Based on the study of small-world basic theory and growing small-world model, traditional high-dimensional indexing technology, the original indexing structure and generation algorithm based on small-world model, this dissertation proposes a novel index structure based on growing small-world network model. Then analyze the basic characteristic parameters, demonstrate the small world properties of the index model. It proposes a new novel generation algorithm and the algorithm has lower time complexity and simpler implementation. The experiment result indicates the index structure and generation algorithm has high efficiency and the query performance is similar to the original index model.(2) Research on updating for high-dimensional indexing technology based on growing network model. Based on the updating algorithm of high-dimensional indexing based on small-world model and the updating algorithm of other high-dimensional indexing. In this dissertation, it proposes an efficient algorithm based on growing small-world model for updating and maintaining the index after inserting and deleting nodes. It includes inserting or deleting one or more nodes. The experiment result demonstrates the small-world property after inserting or deleting nodes.The experiment result shows that the proposed high-dimensional indexing technology based on growing network has better index generation performance than the original indexing technology based on small world model. The updating algorithm ensures that the indexing technology is dynamic.