Personalized new media marketing recommendation system based on TF-IDF algorithm optimizing LSTM-TC model

Personalized recommendation systems can provide users with more targeted and attractive marketing content based on their interests and behavioral characteristics, so it is necessary to explore precise marketing strategies. People are accustomed to using fragmented language on the internet to express their opinions and viewpoints. In order to obtain valuable information from the vast amount of text information, it is necessary to classify these short texts to optimize new media marketing. Therefore, the study introduced word frequency and inverse text frequency for feature weight calculation. At the same time, the study introduced word frequency and inverse text frequency for feature weight calculation, and introduced a classification model combining long short-term memory networks and convolutional neural networks based on text classification, thereby constructing a new media marketing system. The innovation of the research lies in the innovative combination of traditional word frequency and inverse text frequency feature weight calculation methods with deep learning models to accurately capture the importance of text features, thereby improving the performance of classification models. At the same time, combining the advantages of long short-term memory networks and convolutional neural networks, it effectively processes text sequence data, extracts semantic information from it, and achieves more accurate text classification. The outcomes indicated that the effective average F1 value of the classification model combining the short-term memory network and the convolutional neural network based on text classification was 93.26%, the effective average recall rate is as high as 89.11%, and the effective average accuracy was 94.52%. Meanwhile, the minimum effective running time of the model was only 79.83 s. In addition, Ruili Network, which adopted a new media marketing system, had an average of 2.0 daily visits per person, with an effective browsing time of 192.9 s per person, and a daily browsing page count of 7.5 pages per person. This indicates that the new media marketing system has excellent practical application effects and provides reliable technical support for current research in the field of online marketing.