Determinantal Point Process Likelihoods for Sequential Recommendation. Liu, Y., Walder, C., & Xie, L. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, of SIGIR '22, pages 1653–1663, New York, NY, USA, 2022. Association for Computing Machinery. ![Determinantal Point Process Likelihoods for Sequential Recommendation [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Sequential recommendation is a popular task in academic research and close to real-world application scenarios, where the goal is to predict the next action(s) of the user based on his/her previous sequence of actions. In the training process of recommender systems, the loss function plays an essential role in guiding the optimization of recommendation models to generate accurate suggestions for users. However, most existing sequential recommendation tech- niques focus on designing algorithms or neural network architectures, and few efforts have been made to tailor loss functions that fit naturally into the practical application scenario of sequential recommender systems. Ranking-based losses, such as cross-entropy and Bayesian Personalized Ranking (BPR) are widely used in the sequential recommendation area. We argue that such objective functions suffer from two inherent drawbacks: i) the dependencies among elements of a sequence are overlooked in these loss formulations; ii) instead of balancing accuracy (quality) and diversity, only generating accurate results has been over emphasized. We therefore propose two new loss functions based on the Determinantal Point Process (DPP) likelihood, that can be adaptively applied to estimate the subsequent item or items. The DPP-distributed item set captures natural dependencies among temporal actions, and a quality vs. diversity decomposition of the DPP kernel pushes us to go beyond accuracy-oriented loss functions. Experimental results using the proposed loss functions on three real-world datasets show marked improvements over state-of-the-art sequential recommendation methods in both quality and diversity metrics.
@inproceedings{10.1145/3477495.3531965,
author = {Liu, Yuli and Walder, Christian and Xie, Lexing},
title = {Determinantal Point Process Likelihoods for Sequential Recommendation},
year = {2022},
isbn = {9781450387323},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3477495.3531965},
doi = {10.1145/3477495.3531965},
abstract = {Sequential recommendation is a popular task in academic research and close to real-world application scenarios, where the goal is to predict the next action(s) of the user based on his/her previous sequence of actions. In the training process of recommender systems, the loss function plays an essential role in guiding the optimization of recommendation models to generate accurate suggestions for users. However, most existing sequential recommendation tech- niques focus on designing algorithms or neural network architectures, and few efforts have been made to tailor loss functions that fit naturally into the practical application scenario of sequential recommender systems. Ranking-based losses, such as cross-entropy and Bayesian Personalized Ranking (BPR) are widely used in the sequential recommendation area. We argue that such objective functions suffer from two inherent drawbacks: i) the dependencies among elements of a sequence are overlooked in these loss formulations; ii) instead of balancing accuracy (quality) and diversity, only generating accurate results has been over emphasized. We therefore propose two new loss functions based on the Determinantal Point Process (DPP) likelihood, that can be adaptively applied to estimate the subsequent item or items. The DPP-distributed item set captures natural dependencies among temporal actions, and a quality vs. diversity decomposition of the DPP kernel pushes us to go beyond accuracy-oriented loss functions. Experimental results using the proposed loss functions on three real-world datasets show marked improvements over state-of-the-art sequential recommendation methods in both quality and diversity metrics.},
booktitle = {Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval},
pages = {1653–1663},
numpages = {11},
keywords = {sequential recommendation, determinantal point process, loss function, diversity},
location = {Madrid, Spain},
series = {SIGIR '22}
}
Downloads: 0
{"_id":"GSGuPdypL4ZrHYAMN","bibbaseid":"liu-walder-xie-determinantalpointprocesslikelihoodsforsequentialrecommendation-2022","author_short":["Liu, Y.","Walder, C.","Xie, L."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"propositions":[],"lastnames":["Liu"],"firstnames":["Yuli"],"suffixes":[]},{"propositions":[],"lastnames":["Walder"],"firstnames":["Christian"],"suffixes":[]},{"propositions":[],"lastnames":["Xie"],"firstnames":["Lexing"],"suffixes":[]}],"title":"Determinantal Point Process Likelihoods for Sequential Recommendation","year":"2022","isbn":"9781450387323","publisher":"Association for Computing Machinery","address":"New York, NY, USA","url":"https://doi.org/10.1145/3477495.3531965","doi":"10.1145/3477495.3531965","abstract":"Sequential recommendation is a popular task in academic research and close to real-world application scenarios, where the goal is to predict the next action(s) of the user based on his/her previous sequence of actions. In the training process of recommender systems, the loss function plays an essential role in guiding the optimization of recommendation models to generate accurate suggestions for users. However, most existing sequential recommendation tech- niques focus on designing algorithms or neural network architectures, and few efforts have been made to tailor loss functions that fit naturally into the practical application scenario of sequential recommender systems. Ranking-based losses, such as cross-entropy and Bayesian Personalized Ranking (BPR) are widely used in the sequential recommendation area. We argue that such objective functions suffer from two inherent drawbacks: i) the dependencies among elements of a sequence are overlooked in these loss formulations; ii) instead of balancing accuracy (quality) and diversity, only generating accurate results has been over emphasized. We therefore propose two new loss functions based on the Determinantal Point Process (DPP) likelihood, that can be adaptively applied to estimate the subsequent item or items. The DPP-distributed item set captures natural dependencies among temporal actions, and a quality vs. diversity decomposition of the DPP kernel pushes us to go beyond accuracy-oriented loss functions. Experimental results using the proposed loss functions on three real-world datasets show marked improvements over state-of-the-art sequential recommendation methods in both quality and diversity metrics.","booktitle":"Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval","pages":"1653–1663","numpages":"11","keywords":"sequential recommendation, determinantal point process, loss function, diversity","location":"Madrid, Spain","series":"SIGIR '22","bibtex":"@inproceedings{10.1145/3477495.3531965,\nauthor = {Liu, Yuli and Walder, Christian and Xie, Lexing},\ntitle = {Determinantal Point Process Likelihoods for Sequential Recommendation},\nyear = {2022},\nisbn = {9781450387323},\npublisher = {Association for Computing Machinery},\naddress = {New York, NY, USA},\nurl = {https://doi.org/10.1145/3477495.3531965},\ndoi = {10.1145/3477495.3531965},\nabstract = {Sequential recommendation is a popular task in academic research and close to real-world application scenarios, where the goal is to predict the next action(s) of the user based on his/her previous sequence of actions. In the training process of recommender systems, the loss function plays an essential role in guiding the optimization of recommendation models to generate accurate suggestions for users. However, most existing sequential recommendation tech- niques focus on designing algorithms or neural network architectures, and few efforts have been made to tailor loss functions that fit naturally into the practical application scenario of sequential recommender systems. Ranking-based losses, such as cross-entropy and Bayesian Personalized Ranking (BPR) are widely used in the sequential recommendation area. We argue that such objective functions suffer from two inherent drawbacks: i) the dependencies among elements of a sequence are overlooked in these loss formulations; ii) instead of balancing accuracy (quality) and diversity, only generating accurate results has been over emphasized. We therefore propose two new loss functions based on the Determinantal Point Process (DPP) likelihood, that can be adaptively applied to estimate the subsequent item or items. The DPP-distributed item set captures natural dependencies among temporal actions, and a quality vs. diversity decomposition of the DPP kernel pushes us to go beyond accuracy-oriented loss functions. Experimental results using the proposed loss functions on three real-world datasets show marked improvements over state-of-the-art sequential recommendation methods in both quality and diversity metrics.},\nbooktitle = {Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval},\npages = {1653–1663},\nnumpages = {11},\nkeywords = {sequential recommendation, determinantal point process, loss function, diversity},\nlocation = {Madrid, Spain},\nseries = {SIGIR '22}\n}\n\n","author_short":["Liu, Y.","Walder, C.","Xie, L."],"key":"10.1145/3477495.3531965","id":"10.1145/3477495.3531965","bibbaseid":"liu-walder-xie-determinantalpointprocesslikelihoodsforsequentialrecommendation-2022","role":"author","urls":{"Paper":"https://doi.org/10.1145/3477495.3531965"},"keyword":["sequential recommendation","determinantal point process","loss function","diversity"],"metadata":{"authorlinks":{}}},"bibtype":"inproceedings","biburl":"http://cm.cecs.anu.edu.au/documents/publications.bib","dataSources":["CsTNFAMJm4FoD3yFY","8bdqaFbAALQsNy4E3"],"keywords":["sequential recommendation","determinantal point process","loss function","diversity"],"search_terms":["determinantal","point","process","likelihoods","sequential","recommendation","liu","walder","xie"],"title":"Determinantal Point Process Likelihoods for Sequential Recommendation","year":2022}