Neural Session-Aware Recommendation

Phương, Từ Minh; Thanh, Tran Cong; Bach, Ngo Xuan

doi:10.1109/access.2019.2926074

Cited by 28 publications

(9 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Experimental results on two datasets, an e-commerce and a video, shows that their model outperforms existing methods which do not use deep learning. Following this success, other RNN-based models with several modifications and/or extensions such as data augmentation [24], adding attentive mechanism over session actions [17], and combining user identities with session data [19] have been introduced and achieved impressive performance.…”

Section: Deep Learning Based Recommendation Methodsmentioning

confidence: 99%

Dual Transformer Encoders for Session-Based Recommendation

Pham

Bach

Phương

2021

JCC

Self Cite

View full text Add to dashboard Cite

When long-term user proles are not available, session-based recommendation methods are used to predict the user's next actions from anonymous sessions-based data. Recent advances in session-based recommendation highlight the necessity of modeling not only user sequential behaviors but also the user's main interest in a session, while avoiding the eect of unintended clicks causing interest drift of the user. In this work, we propose a Dual Transformer Encoder Recommendation model (DTER) as a solution to address this requirement. The idea is to combine the following recipes: (1) a Transformer-based model with dual encoders capable of modeling both sequential patterns and the main interest of the user in a session; (2) a new recommendation model that is designed for learning richer session contexts by conditioning on all permutations of the session prex. This approach provides a unied framework for leveraging the ability of the Transformer's self-attention mechanism in modeling session sequences while taking into account the user's main interest in the session. We empirically evaluate the proposed method on two benchmark datasets. The results show that DTER outperforms state-of-the-art session-based recommendation methods on common evaluation metrics.

show abstract

Section: Deep Learning Based Recommendation Methodsmentioning

confidence: 99%

Dual Transformer Encoders for Session-Based Recommendation

Pham

Bach

Phương

2021

JCC

Self Cite

View full text Add to dashboard Cite

show abstract

“…It has similar functions to feature vectors. Compared with the computing features of audio, content-based recommendation algorithms have greater advantages in text information and more mature [23]. 3 Mobile Information Systems recommendation algorithm is to mine the user's behavior data to calculate its relevance for recommendation [24].…”

Section: Mobile Information Systemsmentioning

confidence: 99%

Music Recommendation System and Recommendation Model Based on Convolutional Neural Network

Zhang

2022

Mobile Information Systems

View full text Add to dashboard Cite

In today’s era of big data with excess information, music is common and everyday, which shows the huge amount of music data. How to obtain one’s favorite music from the massive music database has become a problem, and the emergence of music recommendation systems is also inevitable. In this paper, we take digital piano music as the research object, form comprehensive features using spectrum and notes, design classification methods using convolutional neural networks, and further process the classification results and design recommendation algorithms. The basic method of music recommendation of this algorithm is to determine the structure of the network model, determine the corresponding training model, and improve the parameters on the basis of the typical source network model used in the system experiment. Historical behavior chooses to collect information. Then, it reads the audio data on the system and retrieves it from Mel, which reveals the identity of the music. The classification proposal achieves its goal by denying the similarity between customer preferences and the potential of two musical characteristics. Two recommended methods based on convolutional neural networks are tested in this article. On the whole, the accuracy of the user’s comprehensive feature, recommendation method is higher than the recommendation accuracy rate of the multicategory user. In the comparison experiment of the single-category and multicategory recommendation methods, the average accuracy rate of single-category user feature recommendation is 50.35%; and the recommendation accuracy rate of multicategory user features is higher than the recommendation accuracy rate of single-category user features. The experimental results show that the two recommendation methods can achieve better recommendation results.

show abstract

“…The number of works that try to incorporate such longer-term preferences in session-aware recommenders (Quadrana et al 2018) is still scarce. Existing works include Quadrana et al (2017) and Phuong et al (2019), but these focus on specific domains, e.g., job or music recommendations. Hence, there is an opportunity to study session-aware recommendations using both on the user's shortterm intents and their longer-term profile.…”

Section: Research Directionsmentioning

confidence: 99%