Generation of sentence representation
摘要
說明書
For example, in the sentence graph 512 illustrated in the examples of
The graph convolution module 526 is configured to apply a graph convolution operation on the set of neighbor nodes to obtain the word representation for the given node. By means of the graph convolution operation, information of the neighbor nodes can be passed to the given node to generate the corresponding word representation. The graph convolution module 526 may be designed to utilize of any convolution operations that can be employed to process graph information. In some embodiments, the graph convolution module 526 may be implemented based on a neural network which can implement representation extraction from a graph. Such neural network may also be referred to as a graph neural network (GNN). The graph convolution module 526 may be implemented as one or more layers in the GNN to perform the graph convolution operation.
