Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations
Book recommendations are crucial in digital library transformation, enhancing service sophistication and customization. They allow readers to access books tailored to their specific interests. In this paper, we propose a novel heterogeneous network embedding approach for personalized book recommenda...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10571949/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1846118366729732096 |
|---|---|
| author | Yafeng Kong |
| author_facet | Yafeng Kong |
| author_sort | Yafeng Kong |
| collection | DOAJ |
| description | Book recommendations are crucial in digital library transformation, enhancing service sophistication and customization. They allow readers to access books tailored to their specific interests. In this paper, we propose a novel heterogeneous network embedding approach for personalized book recommendations. Our model integrates both assessment and representation data within fields. Additionally, it uses a neural network architecture to refine traditional cross-field matrix factorization. By incorporating a nonlinear mapping function, our approach captures field disparities. Furthermore, it also embeds product attribute representations into cross-field recommendations as heterogeneous network embeddings. Consequently, it effectively exploits comprehensive representation data across fields, enhancing book recommendations. The experimental results show that our method achieves RMSE (Root Mean Squared Error) and MAE (Mean Absolute Error) metrics of no higher than 0.767 and 0.605, respectively. These metrics apply across various training set proportions and cold-start customer ratios in both general and customer cold-start scenarios. Compared to other advanced methods, our improvements in RMSE and MAE are not less than 1.01% and 1.13%, respectively. These findings confirm the superiority and robustness of our model in boosting recommendation performance and addressing cold-start issues effectively. |
| format | Article |
| id | doaj-art-a7cf6bbe77104a89a585dac780d8dec9 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-a7cf6bbe77104a89a585dac780d8dec92024-12-18T00:01:17ZengIEEEIEEE Access2169-35362024-01-011218872318873810.1109/ACCESS.2024.341895710571949Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book RecommendationsYafeng Kong0https://orcid.org/0009-0008-0982-8513Jining University, Qufu, Shandong, ChinaBook recommendations are crucial in digital library transformation, enhancing service sophistication and customization. They allow readers to access books tailored to their specific interests. In this paper, we propose a novel heterogeneous network embedding approach for personalized book recommendations. Our model integrates both assessment and representation data within fields. Additionally, it uses a neural network architecture to refine traditional cross-field matrix factorization. By incorporating a nonlinear mapping function, our approach captures field disparities. Furthermore, it also embeds product attribute representations into cross-field recommendations as heterogeneous network embeddings. Consequently, it effectively exploits comprehensive representation data across fields, enhancing book recommendations. The experimental results show that our method achieves RMSE (Root Mean Squared Error) and MAE (Mean Absolute Error) metrics of no higher than 0.767 and 0.605, respectively. These metrics apply across various training set proportions and cold-start customer ratios in both general and customer cold-start scenarios. Compared to other advanced methods, our improvements in RMSE and MAE are not less than 1.01% and 1.13%, respectively. These findings confirm the superiority and robustness of our model in boosting recommendation performance and addressing cold-start issues effectively.https://ieeexplore.ieee.org/document/10571949/Book recommendationdeep learningheterogeneous network embeddinglibrary digitalization |
| spellingShingle | Yafeng Kong Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations IEEE Access Book recommendation deep learning heterogeneous network embedding library digitalization |
| title | Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations |
| title_full | Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations |
| title_fullStr | Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations |
| title_full_unstemmed | Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations |
| title_short | Enhancing Library Digitalization: A Heterogeneous Network Embedding Approach for Personalized Book Recommendations |
| title_sort | enhancing library digitalization a heterogeneous network embedding approach for personalized book recommendations |
| topic | Book recommendation deep learning heterogeneous network embedding library digitalization |
| url | https://ieeexplore.ieee.org/document/10571949/ |
| work_keys_str_mv | AT yafengkong enhancinglibrarydigitalizationaheterogeneousnetworkembeddingapproachforpersonalizedbookrecommendations |