Efficient query processing algorithms for network analysis

Abstract

Users have worked on analyzing various networks to see meaningful implications. Efficient algorithms for important and useful network queries enable such users to quickly get rich information from network data. Thus, it is essential to develop such query processing algorithms for effective network analysis. This dissertation deals with proposing efficient algorithms for several important and useful network queries.

As useful query processing for social network analysis, query processing for influence maximization (IMAX query) is introduced. An IMAX query asks to find k seed nodes which maximize the spread of influence on targets specified in the query. This query can help to understand the relationship between influential users and target users. In this dissertation, first we formulate IMAX query processing, analyze it theoretically, and show the NP-hardness of it. To efficiently answer an IMAX query, we propose an expectation model for the influence spread of a given seed set and a fast greedy-based approximation method using the expectation model. For the expectation model, we exploit a relationship of paths between users in social networks. For the greedy method, we work out an efficient incremental updating of the marginal gain to our objective function. We conduct experiments to evaluate the proposed method with real-life datasets, and compare the results with those of existing methods that are adapted to the problem. From our experimental results, the proposed method is at least an order of magnitude faster than the existing methods in most cases while achieving similar accuracy.

We consider another useful query processing for network analysis, which is a distance sensitivity query. A distance sensitivity query asks to the shortest distance from a node to another node given a set of failed nodes/edges. An efficient query processing algorithm for this query computes the shortest distance in dynamic networks such as road networks and computer networks. I...