Combining different metadata views for better recommendation accuracy
作者:
Highlights:
• Two items representations based on BabelNet concepts extracted from users’ reviews.
• Combination of two item views of features: one statistical and one based on quality.
• Three combination categories: pre, neighborhood and post combination.
• Post combination provides the best results with approach based on machine learning.
• Neighborhood combination has promising results especially with larger neighborhoods.
摘要
•Two items representations based on BabelNet concepts extracted from users’ reviews.•Combination of two item views of features: one statistical and one based on quality.•Three combination categories: pre, neighborhood and post combination.•Post combination provides the best results with approach based on machine learning.•Neighborhood combination has promising results especially with larger neighborhoods.
论文关键词:Recommender systems,Item representation,Unstructured metadata
论文评审过程:Received 24 January 2018, Accepted 18 January 2019, Available online 6 February 2019, Version of Record 13 February 2019.
论文官网地址:https://doi.org/10.1016/j.is.2019.01.008