Tens-embedding: A Tensor-based document embedding method

作者：

Highlights：

• A new unsupervised multi-view document embedding method is proposed.

• Both abstract and specific views of documents are combined with the help of a tensor

• Three different models are proposed for constructing a tensor by these two views.

• Tensor factorization is applied to extract document embeddings.

摘要

•A new unsupervised multi-view document embedding method is proposed.•Both abstract and specific views of documents are combined with the help of a tensor•Three different models are proposed for constructing a tensor by these two views.•Tensor factorization is applied to extract document embeddings.

论文关键词：Natural language processing,Text classification,Text representation,Document embeddings,Tensor factorization,Topic modeling

论文评审过程：Received 18 July 2019, Revised 15 July 2020, Accepted 16 July 2020, Available online 26 July 2020, Version of Record 31 July 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113770