基于投影寻踪降维的文本特征可视化

利用遗传算法优化投影方向，投影寻踪模型将高维的文本特征数据投影到2~3维的低维可视化空间上，并根据高维数据在这个低维空间当中的投影特征值来反映其线性和非线性结构或特征，达到数据降维目的并实现文本数据特征可视化。不仅大大约简了文本挖掘过程的计算复杂性，还有助于在K-means聚类算法中确定初始中心点数目，提高算法精度。实验验证了这种方法应用于文本特征降维的有效性。

Projection-pursuit-based dimension reduction for visualization of text features

Using genetic algorithm to search for the optimal projecting direction, projection pursuit model was used to project text feature data from high-dimensional space into low-dimensional space (2 or 3 dimensions ), and the linear and non-linear structures and features of the high-dimensional data were shown by its projecting feature value in the low dimensional space, therefore dimensionality was reduced and visualization for high-dimensional text feature data was realized. This method is not only cutting down the computing complexity in the process of text mining, but also helping to determine the number of initial center point for K-means algorithm, and improving the accuracy of the algorithm. Experiments demonstrate the efficiency of this method for text feature dimension reduction.