Study On Ontology-Based Scalable Algorithms For Social Network Extraction

There are many large data sets in social network extraction process for some professional domains. When the extraction of social network is applied to a big-scale community, the number of queries to a search engine becomes a key problem. Because search and computation are demanded for all the node pairs, it costs a lot of time and space in large data sets and multi queries. The researchers put forward a calculation method for part of the node pairs, the algorithm scalability problem is proposed in this case. Some methods about the scalability problem can effectively reduce the time or space in the computational process for social network extraction from large data sets, but these methods are also have some problem like veracity is sensitive at the threshold, and so on. In order to enhance efficiency of extraction process, a scalable algorithm called Ontology-Scalability based on ontology is proposed in this thesis. It stores the domain data in ontology and can reduce the queries issued to the research engine so that it can improve the efficiency of the computation and storage for social network extraction.In this thesis we firstly present the definition and relative concepts of social network and the application of social network analysis in the domain of computer science. Then we summarize the existing methods of social network extraction algorithms and simply analyze the advantages and disadvantages of these representative algorithms. We focus on the construction of ontology and present the ontology construction method and put forward a scalable social network extraction algorithm called Ontology-Scalability based on ontology. We particularly describe the idea, implementation process and performance evaluation of the algorithm.The experimental results show that the computation time complexity turns to O(n) from O(n2) and it also saves the storage apace. It has such advantages as using shorter time and getting higher coverage for large data set. Moreover, it uses ontology which is constructed by domain data, so that it can be used in many other systems.