Title: A study of user comment mining and new media marketing under multi-categorical evidence inference

Authors: Yukui Luo

Addresses: Department of Finance and Economics, Yangjiang Polytechnic, Yangjiang, 529566, China

Abstract: We study the construction of a new media sales review mining model based on convolutional neural network (CNN) review text recognition algorithm and artificial intelligence classification decision algorithm of evidence inference. A long and short term memory algorithm is introduced in the CNN model to extract the contextual meaning of the text, and the classification structure of evidence inference is used to carry out the text semantic reclassification of the model. The experimental results show that the support of evidence classification of the model is within the range of 0.77-0.79, and the correlation between evidence and classification is obvious. The classification accuracy is 0.947 and the recall rate is 0.728. Compared with other experimental algorithms, the classification accuracy and recall rate of the proposed algorithm are higher. The correlation experiments and validity experimental data prove that the CNN-ER model's has high classification accuracy, stable algorithm computation and robustness in online sales user evaluation.

Keywords: CNN; convolutional neural network; evidence inference algorithm; semantic recognition; network sales; long and short-term memory algorithm; classification support.

DOI: 10.1504/IJISD.2024.140844

International Journal of Innovation and Sustainable Development, 2024 Vol.18 No.5/6, pp.745 - 760

Received: 24 Jun 2022
Accepted: 25 Oct 2022
Published online: 03 Sep 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article