SRU-based Multi-angle Enhanced Network for Semantic Text Similarity Calculation of Big Data Language Model

As a fundamental problem of natural language processing (NLP), the calculation semantic text similarity plays crucial role in variety big data application situations. In process modeling, however, owing to complexity and ambiguity Chinese semantics, effectively capturing interaction characteristics only from single angle is impossible. This study proposes deep learning-based computational model for called SRU-based multi-angle enhanced network (SMAEN). Specifically, authors firstly combine character-grained embeddings word-granularity obtained pre-trained represent text. The encoded using bidirectional simple recurrent unit (Bi-SRU) network, local represented soft-aligned attention technique. addition, integrate Bi-SRU with an improved convolutional neural (CNN) global modeling capture semantic, time, spatial short interaction. Finally, they employ pooling layer aggregate results into fixed-length vector multi-layer perceptual (MLP) classifier make determination. Experimental on public datasets LCQMC PAWS-X show that proposed method fully captures features multiple angles achieves advanced performance. can produce better matching enhance accuracy large analysis. It applicable numerous scenarios involving data, such as information retrieval recommendation systems.