Research on uncertainty inference of associate projects based on the latent variable model

To address the problem of data and model uncertainty of reasoning the potential association between projects through text mining, an uncertainty reasoning model combining Latent Variable Model (LVM) and Bayesian Neural Networks (BNNs) is proposed. Firstly, the source of uncertainty in reasoning the relationship between projects without labeled text data is analyzed, and a latent variable model of uncertainty inference is established. Then, according to the Bayesian inference technique, the posterior distribution of latent variables is calculated by the variational inference method, and the optimization objective function is determined. Secondly, based on the graph attention mechanism, a text feature extraction method combining text graph modeling and BNNs is proposed, and the text feature representation is used for classification prediction. In terms of performance evaluation, the experiments on the benchmark dataset show that the classification performance of the model is better than that of the baseline model, and the F1 value index of the two topics is improved by more than 10%. Finally, the effectiveness of this method is verified by an example analysis of the project text dataset.