Aiming at the problems of entity nesting and relationship overlap in the process of tourism domain information extraction, a joint entity relationship extraction method based on multi-dimensional features and mixed time series was proposed. The relationship extraction model based on multi-dimensional features and mixed time series is constructed by coding layer, entity extraction layer and relationship extraction layer. In the coding layer, BiLSTM can better capture the bidirectional semantic dependence, and use the bidirectional long and short memory network to encode sentences in combination with word context information. Entity extraction layer uses CNN to classify annotated text data, relationship extraction layer integrates the representation ability of different dimensions for relationship extraction, and uses weighted dot product to calculate the loss of relationship extraction to improve the effect of relationship extraction. The experimental results show that the recall rate of the model in this paper is 97.46%, the accuracy is 97.18%, and the F1 value is close to 1. When the threshold is 0.41, the F1 value is 0.805, and the curve reaches the inflection point, then the model has the best effect. The problem of entity nesting and relationship overlap is solved effectively.