Social content characterization method and system fusing text and label network

The invention discloses a social content characterization method and system fusing a text and a label network, and the method comprises the steps: carrying out the processing of text data, obtaining the graph representation of the text data, and carrying out the sorting of nodes in an obtained sub-graph according to a BFS; carrying out data structuring processing on the graph representation; establishing a label network, and randomly walking in the label network according to the meta-path to obtain vector representation of the label; and inputting the processed structured data into a neural network, and obtaining vector representation of a label based on the label network and the meta path to carry out LSTM + RNN training on the neural network. 本申请公开了一种融合文本和标签网络的社交内容表征方法和系统，所述方法包括：对文本数据进行处理，得到该文本数据的图表示，并对得到的子图中节点按照BFS进行排序；对所述图表示进行数据结构化处理；建立标签网络，并根据元路径在标签网络中随机游走得到标签的向量表示；将处理得到的结构化数据输入神经网络中，基于标签网络和元路径得到标签的向量表示对所述神经网络进行LSTM+RNN训练。