TY - GEN
T1 - Secure Analysis of Social Media Data
AU - Siddaramappa, Hareesha Katiganere
AU - Maradithaya, Sumana
AU - Kumar, Sampath
PY - 2019/4/1
Y1 - 2019/4/1
N2 - Confidentiality of the social media data during analysis is a major concern. Several real evidences show how the privacy and security of the data is compromised. One of the essential processes with social media data is to find the shortest paths between selected pair of nodes. This paper proposes a technique to modify the original data before analysis. The algorithm calculates shortest paths (data utility) between target nodes and then classifies edges into partially visited, all-visited and unvisited edges. Each category of edges is then perturbed using a dynamic variable value that is bound to satisfy specific constraints such that the shortest path as well as the shortest paths lengths, between the target node pairs remains the same. This paper proposes an approach to preserve the privacy of the weights and also generates an accurate length of the shortest path. It is also observed that the shortest path lengths between any target pairs of nodes are retained. The output is in the form of graphs, that shows that the proposed perturbation strategy perturbs the sensitive edge weights up to a maximum 72%, while keeping the difference in shortest path lengths minimum (up to 3%). It is hence demonstrated that along with preserving the sensitive information by perturbing the edge weights, the data utility is preserved i.e. the shortest path lengths are kept as near as potential to the original ones.
AB - Confidentiality of the social media data during analysis is a major concern. Several real evidences show how the privacy and security of the data is compromised. One of the essential processes with social media data is to find the shortest paths between selected pair of nodes. This paper proposes a technique to modify the original data before analysis. The algorithm calculates shortest paths (data utility) between target nodes and then classifies edges into partially visited, all-visited and unvisited edges. Each category of edges is then perturbed using a dynamic variable value that is bound to satisfy specific constraints such that the shortest path as well as the shortest paths lengths, between the target node pairs remains the same. This paper proposes an approach to preserve the privacy of the weights and also generates an accurate length of the shortest path. It is also observed that the shortest path lengths between any target pairs of nodes are retained. The output is in the form of graphs, that shows that the proposed perturbation strategy perturbs the sensitive edge weights up to a maximum 72%, while keeping the difference in shortest path lengths minimum (up to 3%). It is hence demonstrated that along with preserving the sensitive information by perturbing the edge weights, the data utility is preserved i.e. the shortest path lengths are kept as near as potential to the original ones.
UR - http://www.scopus.com/inward/record.url?scp=85070640808&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85070640808&partnerID=8YFLogxK
U2 - 10.1109/ICACTM.2019.8776834
DO - 10.1109/ICACTM.2019.8776834
M3 - Conference contribution
T3 - 2019 International Conference on Automation, Computational and Technology Management, ICACTM 2019
SP - 315
EP - 319
BT - 2019 International Conference on Automation, Computational and Technology Management, ICACTM 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2019 International Conference on Automation, Computational and Technology Management, ICACTM 2019
Y2 - 24 April 2019 through 26 April 2019
ER -