A Knowledge–Enhanced Text Clustering Based Adversarial Learning for Text Generation

Aiming at the problem that users find it difficult to quickly filter out effective information when faced with massive amounts of data. The advantage of using a generative adversarial network mod el is that it does not require supervision and data labeling. Traditional generative adversarial network text generation models are prone to mode collapse. A knowledge-enhanced cluster text generation adversarial network model is proposed. This model improves the model structure, designs an auxiliary training module, combines the clustering algorithm of knowledge learning, integrates the generative adversarial network and the clustering algorithm to generate text, and adjusts the generative network by training the results generated by the discriminator. And parameters in clustering algorithms to generate more diverse, high-quality text. In this article, we verified the text under two different data sets. Compared with the evaluation indicators of other models in the same environment, the BLUE index of this model increased by 0.9% ~ 19.8%. The evaluation index is better than the baseline model, the results show that the model has high feasibility and effectiveness.