openhgnn.models.MeiREC

class MeiREC(config)[source]

MeiREC from paper Metapath-guided Heterogeneous Graph Neural Network for Intent Recommendation in KDD_2019.

Code from author.

We leverage metapaths to obtain different-step neighbors of an object, and the embeddings of us ers and queries are the aggregation of their neighbors under different metapaths.And we propose to represent the queries and items with a small number of term embeddings.we need to learn the term embeddings, rather than all object embeddings. This method is able to significantly reduc e the number of parameters.

Parameters:

  • user_seq_length (int) – Number for process dataset.

  • ... –

  • batch_num (int) – Number of batch.

  • weight_decay (float) – Number of weight_decay.

  • lr (float) – learning rate.

  • train_epochs (int) – Number of train epoch.

  • ----------- –