Fast mining of relationships of large-scale data

As a result of a deployment of ICT, large-sized data is available these days. Since graphical structures can effectively capture sparse relationships among data, they are actively studied recently. In this study, we developed an efficient algorithm to construct L1-graph from large size of data. We reduce the computation cost by analyzing conditions with which nodes have an edge between them. Since our algorithm significantly improves the computation speed comparing to existing approaches, it can efficiently reveal relationships behind data. This indicates that we can effectively analyze large-sized data which is not considered to apply graph mining approaches. We can effectively perform recommendation and prediction based on big data. In the near future, we can obtain world-wide data as by using IoT. We believe our algorithm is useful in analyzing such data.

Please click the thumbnail image to open the full-size PDF file.