Abstract
Drastic increase in item/product review volume has caused serious information overload. Traditionally, product reviews are exhibited in chronological or popularity order without personalization. The review recommendation model provides users with attractive reviews and efficient consumption experience, allowing users to grasp the characteristics of items in seconds. However, the sparsity of interactions between users and reviews appears to be a major challenge. And the multi-relationship context, especially potential semantic feature in reviews, is not fully exploited. To address these problems, Multi-Task Learning model incorporating Personalized Transformer (MTL-PT) is proposed to provide users with an interesting review list. It contains three tasks: the main task models user’s preference to reviews with the proposed poly aggregator, incorporating the user-item-aware semantic feature. Two auxiliary tasks model the quality of reviews, and user-item interactions, respectively. These tasks collaboratively learn the multi-relationship among user, item, and review. The shared latent features of user/item link the three tasks together. Especially, the personalized semantic features of reviews are also fused into the tasks with the proposed personalized transformer. Two new real-world datasets for personalized review recommendation are collected and constructed. Extensive experiments are conducted on them. Compared with the state-of-the-arts, the results validate the effectiveness of our model for review recommendation.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
Since the inputs of personalized transformers in the user-review task and the item-review task are similar, here we only take the personalized transformer input in the user-review task as an example.
- 2.
- 3.
If user replies a review, we assume user is interested in the review, otherwise not.
- 4.
The datasets and code would be public later.
- 5.
- 6.
References
Agarwal, D., Chen, B., Pang, B.: Personalized recommendation of user comments via factor models. In: EMNLP, pp. 571–582 (2011)
Chen, C., Yang, Y., Zhou, J., Li, X., Bao, F.S.: Cross-domain review helpfulness prediction based on convolutional neural networks with auxiliary domain discriminators. In: NAACL, pp. 602–607 (2018)
Devlin, J., Chang, M., Lee, K., Toutanova, K.: BERT: pre-training of deep bidirectional transformers for language understanding. In: NAACL, pp. 4171–4186 (2019)
Diaz, G.O., Ng, V.: Modeling and prediction of online product review helpfulness: a survey. In: ACL, pp. 698–708 (2018)
Du, J., Zheng, L., He, J., Rong, J., Wang, H., Zhang, Y.: An interactive network for end-to-end review helpfulness modeling. Data Sci. Eng. 5(3), 261–279 (2020)
Fan, M., Feng, C., Guo, L., Sun, M., Li, P.: Product-aware helpfulness prediction of online reviews. In: WWW, pp. 2715–2721 (2019)
He, X., Liao, L., Zhang, H., Nie, L., Hu, X., Chua, T.: Neural collaborative filtering. In: WWW, pp. 173–182 (2017)
Huang, C., Jiang, W., Wu, J., Wang, G.: Personalized review recommendation based on users’ aspect sentiment. TOIT 20(4), 1–26 (2020)
Humeau, S., Shuster, K., Lachaux, M., Weston, J.: Poly-encoders: architectures and pre-training strategies for fast and accurate multi-sentence scoring. In: ICLR (2020)
Lu, Y., Tsaparas, P., Ntoulas, A., Polanyi, L.: Exploiting social context for review quality prediction. In: WWW, pp. 691–700 (2010)
Ma, J., Wen, J., Zhong, M., Chen, W., Li, X.: MMM: multi-source multi-net micro-video recommendation with clustered hidden item representation learning. Data Sci. Eng. 4(3), 240–253 (2019)
Martin, L., Pu, P.: Prediction of helpful reviews using emotions extraction. In: AAAI, pp. 1551–1557 (2014)
Moghaddam, S., Jamali, M., Ester, M.: Review recommendation: personalized prediction of the quality of online reviews. In: CIKM, pp. 2249–2252 (2011)
Moghaddam, S., Jamali, M., Ester, M.: ETF: extended tensor factorization model for personalizing prediction of review helpfulness. In: WSDM, pp. 163–172 (2012)
Momeni, E., Cardie, C., Diakopoulos, N.: A survey on assessment and ranking methodologies for user-generated content on the web. ACM Comput. Surv. 48(3), 41:1–41:49 (2016)
Paul, D., Sarkar, S., Chelliah, M., Kalyan, C., Nadkarni, P.P.S.: Recommendation of high quality representative reviews in e-commerce. In: RecSys, pp. 311–315 (2017)
Prado, T.R.P., Moro, M.M.: Review recommendation for points of interest’s owners. In: HT, pp. 295–304 (2017)
Rendle, S., Freudenthaler, C., Gantner, Z., Schmidt-Thieme, L.: BPR: Bayesian personalized ranking from implicit feedback. CoRR arXiv:1205.2618 (2012)
Suresh, V., Roohi, S., Eirinaki, M., Varlamis, I.: Using social data for personalizing review rankings. In: RecSys (2014)
Tang, J., Gao, H., Hu, X., Liu, H.: Context-aware review helpfulness rating prediction. In: RecSys, pp. 1–8 (2013)
Tang, J., Gao, H., Liu, H.: mTrust: discerning multi-faceted trust in a connected world. In: WSDM, pp. 93–102 (2012)
Vaswani, A., et al.: Attention is all you need. In: NIPS, pp. 5998–6008 (2017)
Wang, C., Chen, G., Wei, Q.: A temporal consistency method for online review ranking. Knowl. Based Syst. 143, 259–270 (2018)
Wang, S., Li, C., Zhao, K., Chen, H.: Context-aware recommendations with random partition factorization machines. Data Sci. Eng. 2(2), 125–135 (2017)
Wu, C., Wu, F., An, M., Huang, J., Huang, Y., Xie, X.: Neural news recommendation with attentive multi-view learning. In: IJCAI, pp. 3863–3869 (2019)
Wu, C., Wu, F., An, M., Huang, J., Huang, Y., Xie, X.: NPA: neural news recommendation with personalized attention. In: SIGKDD, pp. 2576–2584 (2019)
Xiong, W., Litman, D.J.: Empirical analysis of exploiting review helpfulness for extractive summarization of online reviews. In: COLING, pp. 1985–1995 (2014)
Acknowledgments
This work is supported by the National Natural Science Foundation of China (61902439, U19112031), Guangdong Basic and Applied Basic Research Foundation (2021A1515011902, 2019A1515011159), National Science Foundation for Post-Doctoral Scientists of China under Grant (2019M663237), Macao Young Scholars Program (UMMTP2020-MYSP-016), the Major Science and Technology Research Programs of Zhongshan City (2019B2006, 2019A40027, 2021A1003).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Wang, H., Liu, W., Yin, J. (2021). Multi-Task Learning with Personalized Transformer for Review Recommendation. In: Zhang, W., Zou, L., Maamar, Z., Chen, L. (eds) Web Information Systems Engineering – WISE 2021. WISE 2021. Lecture Notes in Computer Science(), vol 13081. Springer, Cham. https://doi.org/10.1007/978-3-030-91560-5_12
Download citation
DOI: https://doi.org/10.1007/978-3-030-91560-5_12
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-91559-9
Online ISBN: 978-3-030-91560-5
eBook Packages: Computer ScienceComputer Science (R0)