Fake Comment Detection Based on Time Series and Density Peaks Clustering.

This paper proposes a fake comment recognition method based on time series and density peaks clustering. Firstly, an anomaly recognition model based on multi-dimensional time series is constructed. Secondly, according to the idea of multi-scale features, seven benchmark-scale and corresponding subdivision-scale features are extracted hierarchically, and further, 49 features are finally obtained. At last, an optimized detection model based on density peaks clustering is proposed for identifying the fake comments, so as to improve the anti-noise ability of our method. The effectiveness of our proposed method is verified by several experiments, with the AUC value reaching 92%.