Improved Convolutional Neural Network Algorithm for Product Text Subject and Rights Protection Research Recognition Algorithm

Abstract:
Aiming at the problems of poor text feature representation ability and low accuracy of model topic recognition in traditional methods, a text topic recognition method integrating SENet and convolutional neural network is proposed. Integrate the Word2vec word vector corresponding to each word with the LDA topic vector, and process the document weighted vectorization of the topic of the e-commerce product detail page according to the word; build the SECNN topic recognition model, and use SENet to perform the feature map output from the convolution layer. Re-calibration of weights, relying on it to improve the performance of important features and suppress useless features, efficiently carry out topic identification; use FDA to evaluate the class representation ability of samples, propose FDA-SGD algorithm to optimize model parameters, and complete e-commerce product details Page text topic recognition task. The effectiveness of the improved algorithm is verified by using the news text data set, and the comparison with the traditional model shows that the improved algorithm can effectively improve the convergence speed of the model and has better topic recognition ability.