Abstract:

Social network data acquisition has become a foundation of social network analysis. Although the majority of social media provides official APIs for the developers, there are always rigid limitations on frequency of calls, permissions and contents, which makes it difficult to acquire complete data. So data acquisition technologies based on user login simulation are particularly important. Nevertheless, the login process of the majority of social media has some inherent security risks. The login passwords are usually transmitted as clear text, and user privacy suffers serious threats. This paper gives a detailed analysis of the interactions between clients and servers during user login based on Twitter. When we parse POST requests on the network traffic level, we find that Twitter transmits login passwords as clear text. Hence, we propose an encryption algorithm of double chaos with mutual feedback based on the features of social networks. The algorithm utilizes users’ IDs, the creation time and the number of followings as the initial values of the encryption functions. By applying interactive operations between Logistic mapping and Tent mapping, the algorithm obtains a new sequence of secret keys. The particularity of input parameters leads to the unpredictability of the cipher text. Experimental results show that our algorithm obtains better encryption and decryption. Meanwhile, the speeds of encryption and decryption are on the time scale of milliseconds, thus users do not feel any waiting. Furthermore, the algorithm has the features of sensitivity to initial conditions, large key space and highstrength encryption. The algorithm can prevent attackers from breaking the codes via methods such as phase diagram, bruteforce computation and statistics, and has a wide application prospect.

Key words: social network;login simulation;chaotic encryption