Enhancing Enterprise Sentiment Analysis With Word2vec-Bi-Gru-Attention Method

In business management, the use of natural language processing for public opinion analysis has become an important strategic tool. However, traditional text analysis methods, such as simple machine learning models and rudimentary deep learning networks, often suffer from limitations in analysis accuracy and efficiency due to the loss of contextual information and insufficient ability to handle long-distance dependencies. In response to these problems, this article proposes a composite model called “Word2Vec-Bi-GRU-Attention” that integrates Word2Vec, Bidirectional Gated Recurrent Unit (Bi-GRU) and attention mechanism. The model optimizes the vector representation of words through Word2Vec, uses Bi-GRU to capture contextual information in the text, and uses the attention mechanism to strengthen the model's ability to focus on key information, thereby improving the accuracy and detail of the analysis. We conducted extensive testing on a homemade “CorpSurveyNet” dataset, which contains corporate opinion data from multiple industries. Experimental results show that our method improves accuracy by 12 % over the traditional Bi-GRU model, and improves recall rate and F1 score by 15 % and 14 % respectively. The research in this article not only points out the shortcomings of traditional methods, but also provides a more efficient and accurate technical path for corporate public opinion analysis.