Research and Application Based on Improved Call Fingerprint Identification Algorithm

In order to effectively determine whether a user is a re-entry user, this paper is based on an unproved call fingerprinting algorithm. Research and analysis were carried out by mining relevant characteristics of new and old users' contact circles (including call characteristics and SMS features). On the basis of the weighting algorithm TF-IDF in the call fingerprint algorithm, the speech and audio tunes were combined with the short message frequency. At the same time, on the basis of the speech and audio times, the parameter of the length of the call was increased to further reflect the weight of the user's communication circle number. The improved call fingerprint recognition algorithm can improve the accuracy of identifying re-entry users, and can also provide a new theoretical basis for the discrimination o f multi-card users. The Experiments showed that the model could effectively determine whether a user was re-entering the network and had a high detection accuracy.