Designing an intelligent push model for user emotional topics based on dynamic text categorization in social media news dissemination

Abstract

The exploration of social media comment analysis has garnered considerable scholarly attention in recent epochs, precipitated by the pervasive ubiquity of social media platforms and the copious volume of commentaries engendered by their users. As the prevalence of users disseminating opinions, engaging in news discourse, and articulating sentiments on social media escalates, scrutinizing social media comments assumes paramount significance. This treatise employs a sophisticated deep network model for sentiment classification predicated on online social media textual commentary data, utilizing a bidirectional long short-term memory (BI-LSTM) network. The model initiates data input processing by employing word segmentation and word vector extraction, culminating in the formation of an attention-based bidirectional long short-term memory (ATT-Bi-LSTM) model, which incorporates an attention mechanism for discerning positive and negative emotions. Notably, the model attains recognition rates exceeding 80% for both categories of emotions within the public dataset. Concurrently, the model undergoes training migration for practical application validation using the public dataset, yielding recognition accuracy surpassing 90% in authentic testing scenarios. This substantiates the efficacy of the proposed methodology in proficiently accomplishing the emotion classification task within the dynamic text milieu of social media news propagation. Such proficiency, in turn, furnishes pivotal technical underpinnings for subsequent iterations of intelligent push models and astute public opinion analyses.