Research on the Effectiveness Identification of Novel Reader Comments Based on BERT-Attention Mechanism

Abstract

With the rapid rise of generative artificial intelligence (AI) in the field of literature, online literature reviews have surged, but issues such as difficulty in filtering effective information and inconsistent standards have emerged, posing challenges for reader feedback extraction and content optimisation. Accurately identifying the core valuable information in novel reviews is of significant importance for the construction of the online literature industry ecosystem. This paper focuses on the specific scenario of novel reader reviews and explores the application of the Bidirectional Encoder Representations from Transformers (BERT)-attention mechanism in identifying their effectiveness, using literature research, inductive and deductive methods, and comparative review methods to provide an overview. The study clarifies the criteria for defining effective information in novel reviews, analyses the contextual adaptation bottlenecks of existing technologies, and ultimately constructs a theoretical framework suited for this scenario, providing theoretical support for the implementation of review analysis technology.