Unsupervised Learning of Paragraph Embeddings for Context-Aware Recommendation

Jin Xie,Minxue Huang,Fuxi Zhu,Naixue Xiong,Sheng Huang,Wei Xiong

doi:10.1109/access.2019.2906659

Jin Xie, Minxue Huang + Show 4 more

Open Access

PDF Available

https://doi.org/10.1109/access.2019.2906659

Copy DOI

Export

Save

Cite

Abstract
Highlights/Summary
Full-Text PDF
Similar Papers

Abstract

Listen

The sparsity of data is one of the main reasons restricting the performance of recommender systems. In order to solve the sparsity problem, some recommender systems use auxiliary information, especially text information, as a supplement to increase the prediction accuracy of the ratings. However, the two mainstream approaches based on text analysis have some limitations. The bag-of-words-based model is one of them, being difficult to use the contextual information of the paragraph effectively so that only the shallow understanding of paragraph can be parsed. Another model based on deep learning can extract the contextual information of the paragraph, but it also increases the complexity of the model. This paper proposes a novel context-aware recommendation model named paragraph vector matrix factorization (P2VMF) which integrates the unsupervised learning of paragraph embeddings into probabilistic matrix factorization (PMF). Therefore, P2VMF can capture the semantic information of the paragraph and can improve the prediction accuracy of the ratings. Our extensive experiments on real-world datasets show that the performance of the P2VMF model is preferable as compared with those multiple recommendation models in the situation, where the ratings are quite sparse. And we also verified that the P2V part of the model can well express the semantics in the form of vectors.

Highlights

Context-Aware Recommender Systems (CARS) [1] has been introduced as a way to solve the data sparsity and cold start problems in the recommender systems, and proved to be providing recommendation lists with high accuracy and user satisfaction [2], [3]
We propose a series of context-aware recommendation models named paragraph vector matrix factorization (P2VMF), which capture the contextual information by utilizing paragraph vector (P2V) from the item description document and further enhance the rating prediction accuracy
The main contributions of this study are as follows: (1) We extend the CBOW approach to get a context-aware recommendation model based on the paragraph vector (P2VMF). (2) In order to effectively utilize the semantic of item description and rating data, we combine P2V with probabilistic matrix factorization (PMF) seamlessly

Summary

Introduction

Context-Aware Recommender Systems (CARS) [1] has been introduced as a way to solve the data sparsity and cold start problems in the recommender systems, and proved to be providing recommendation lists with high accuracy and user satisfaction [2], [3]. Among various types of auxiliary information, item description can express the user’s view directly which can’t be translated by computer. In Natural Language Processing field, experts attempt to explain the semantics of users’ view on the documents by using models such as Latent Dirichlet Allocation (LDA) [4], which is incorporated into recommender systems [5]–[7]. As a tool of semantic analysis, LDA classifies words with similar latent semantics by topics. For contextual information of documents, LDA has great limitations. The one reason is that LDA as a topic

Methods

Results

Conclusion