Research on Undesirable text Recognition based on Combination kernel function

摘要

Aiming at the specific application of undesirable text information identification, combining traditional kernel function the paper structured a new combination kernel. This combination kernel builds model for independent characteristics using the linear kernel function, and combined by polynomial kernel function and linear kernel function. Balancing collinear information of individual characteristics and the single characteristic vocabulary and the multiple characteristic vocabulary to determine undesirable properties of the document, it can accurately distinguish some boundary samples. And the recognition accuracy rate is greater than other kernel functions for the application of undesirable text information filtering. Especially the Rcall value achieved satisfactory results.