Rights statement: This is the author’s version of a work that was accepted for publication in Expert Systems with Applications. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Expert Systems with Applications, 168, 2021 DOI: 10.1016/j.eswa.2020.114469
Accepted author manuscript, 2.63 MB, PDF document
Available under license: CC BY-NC-ND
Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Article number | 114469 |
---|---|
<mark>Journal publication date</mark> | 15/04/2021 |
<mark>Journal</mark> | Expert Systems with Applications |
Volume | 168 |
Number of pages | 8 |
Publication Status | Published |
Early online date | 13/12/20 |
<mark>Original language</mark> | English |
Personalized recommendation is a popular research direction in both industry and academia. Some research on recommender systems utilizes the users’ interaction history on items to represent the users’ interests, which has achieved remarkable success. Users’ interests in the real world are dynamically changing and have a strong correlation with the interaction sequence. However, sometimes users’ interests are less relevant to the order of the current interaction sequence, but are more relevant to certain items in the user interaction history. In this paper, a novel deep neural network model is proposed to deal with this situation. The developed model consists of two parts: the present interest relevant to the order of the interaction sequence and the comprehensive interest relevant to some items in the interaction sequence. An ancillary multi-layer perceptron (MLP) is constructed to improve the training of our model. Experiments on public and industrial datasets are conducted. The experimental results show that our proposed model outperforms the state-of-the-art models which demonstrates the effectiveness of the ancillary MLP.