Доклады Российской академии наук. Математика, информатика, процессы управления, 2023, T. 514, № 2, стр. 395-416
НЕ БЫВАЕТ ДВУХ ОДИНАКОВЫХ ПОЛЬЗОВАТЕЛЕЙ: НЕЙРОСЕТЕВАЯ КЛАСТЕРИЗАЦИЯ НА ОСНОВЕ ПОСЛЕДОВАТЕЛЬНОСТЕЙ СОБЫТИЙ ДЛЯ ГЕНЕРАЦИИ АУДИТОРИЙ
В. Жужель 1, *, В. Грабарь 1, Н. Каплоухая 4, Р. Ривера-Кастро 2, 3, 4, Л. Миронова 1, А. Зайцев 1, Е. Бурнаев 1
1 Сколковский институт науки и технологий
121205 Москва, Россия
2 Choco Communications
10967 Берлин, Hasenheide 54, Германия
3 Центр цифровых технологий и управления
80333 Мюнхен,
Arcisstr. 21, Германия
4 Исследования, проведенные в период работы
в Сколковском институте науки и техники
121205 Москва, Россия
* E-mail: vladislav.zhuzhel@skoltech.ru
Поступила в редакцию 01.09.2023
После доработки 15.09.2023
Принята к публикации 18.10.2023
- EDN: CWIIDI
- DOI: 10.31857/S2686954323601859
Полные тексты статей выпуска доступны в ознакомительном режиме только авторизованным пользователям.
Аннотация
Определение нужного пользователя для таргетинга является общей задачей для различных интернет-платформ. Хотя многие системы решают ее, они в значительной степени адаптированы к конкретным особенностям. Из-за этого на практике становится непросто применить данные задачи. Причина в том, что большинство систем предназначены для работы с миллионами активных пользователей и с личной информацией, как в случае с социальными сетями или другими сервисами с высокой виральностью. В литературе мало представлены решения, которые предназначены для обработки данных среднего размера, где единственными доступными данными являются последовательности событий пользователя. Это мотивирует нас представить Look-A-Liker (LAL) как систему глубокой кластеризации. Он использует временные точечные процессы для идентификации похожих пользователей для решения задач таргетинга. Для экспериментов мы используем данные ведущего интернет-маркетплейса гастрономического сектора. LAL обобщает не только закрытые данные. Используя последовательности событий пользователей, можно получить результаты мирового уровня, сравнимые с результатами, получаемыми с использованием новых методов, таких как трансформеры и мультимодальное обучение. Наш подход позволяет повысить оценку по метрике ROC AUC до 20% на реальных наборах данных с 0.803 до 0.959. Хотя LAL фокусируется на сотнях тысяч последовательностей, мы показываем, что его можно применить и в задачах с миллионами пользовательских последовательностей. Мы предоставляем полностью воспроизводимую реализацию с кодом и наборами данных в https://github.com/adasegroup/sequence_clusterers.
Полные тексты статей выпуска доступны в ознакомительном режиме только авторизованным пользователям.
Список литературы
Mathilde Caron et al. “Unsupervised learning of visual features by contrasting cluster assignments”. B: Advances in Neural Information Processing Systems. 2020. V. 33. P. 9912–9924.
Gromit Yeuk-Yin Chan et al. “Interactive Audience Expansion On Large Scale Online Visitor Data”. B: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. New York, NY, USA: Association for Computing Machinery, aвг. 2021. P. 2621–2631.
Daley D., Vere-Jones D. “An introduction to the theory of point processes. Vol. I: Elementary theory and methods. 2nd ed”. B: Vol. 1 (янв. 2003). https://doi.org/10.1007/b97277
Arthur P. Dempster, NanMLaird, Donald B. Rubin. “Maximum likelihood from incomplete data via the EM algorithm”. B: Journal of the Royal Statistical Society: Series B (Methodological). 1977. V. 39.1. P. 1–22.
Stephanie deWet, Jiafan Ou. “Finding Users Who Act Alike: Transfer Learning for Expanding Advertiser Audiences”. B: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. KDD ’19. Anchorage, AK, USA: Association for Computing Machinery, июль 2019. P. 2251–2259.
Dixin Luo et al. “You Are What You Watch and When You Watch: Inferring Household Structures From IPTV Viewing Data”. B: IEEE Transactions on Broadcasting, vol. 60, no. 1, pp. 61–72, March 2014. 2014.
Khoa D. Doan, Pranjul Yadav, Chandan K Reddy. “Adversarial Factorization Autoencoder for Look-alike Modeling”. B: Proceedings of the 28th ACM International Conference on Information and Knowledge Management. CIKM ’19. Beijing, China: Association for Computing Machinery, нoяб. 2019. P. 2803–2812.
Faraone M.F. et al. “Using context to improve the effectiveness of segmentation and targeting in e-commerce”. B: Expert Systems with Applications. 2012. V. 39.9. P. 8439–8451. issn: 0957-4174. https://doi.org/10.1016/j.eswa.2012.01.174. url: https://www.sciencedirect.com/science/article/pii/S0957417412002023.
Fursov I. et al. “Gradient-based adversarial attacks on categorical sequence models via traversing an embedded world”. B: AIST. 2020.
Fursov Ivan et al. “Sequence Embeddings Help Detect Insurance Fraud”. B: IEEE Access. 2022. V. 10. P. 32060–32074.
Felix A. Gers, Jürgen Schmidhuber, Fred Cummins. “Learning to forget: Continual prediction with LSTM”. B. 1999.
Jean-Bastien Grill et al. “Bootstrap your own latent-a new approach to self-supervised learning”. B: Advances in Neural Information Processing Systems. 2020. V. 33. P. 21271–21284.
Luo Haiyan et al. Methods and systems for near real-time lookalike audience expansion in ads targeting. en. https://patents.justia.com/patent/10853847. Accessed: 2022-2-7.
Alan G. Hawkes. “Spectra of some self-exciting and mutually exciting point processes”. B: Biometrika. 1971. V. 58.1. P. 83–90.
Kaiming He et al. “Masked autoencoders are scalable vision learners”. B: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022. P. 16000–16009.
Sergey Ioffe, Christian Szegedy. “Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift”. B: (фeвp. 2015).
Ashish Jaiswal et al. “A survey on contrastive self-supervised learning”. B: Technologies. 2020. V. 9.1. P. 2.
Jinling Jiang et al. Comprehensive audience expansion based on end-toend neural prediction. http://ceur-ws.org/Vol-2410/paper12.pdf. Accessed: 2022-2-8.
Yoon Kim. “Convolutional Neural Networks for Sentence Classification”. B: EMNLP. 2014.
Dennis Koehn, Stefan Lessmann, Markus Schaal. “Predicting online shopping behaviour from clickstream data using deep learning”. B: Expert Systems with Applications. 2020. V. 150. P. 113342. issn: 0957-4174. https://doi.org/10.1016/j.eswa.2020.113342. url: https://www.sciencedirect.com/science/article/pii/S0957417420301676.
Haishan Liu et al. “Audience Expansion for Online Social Network Advertising”. B: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2016.
Haishan Liu et al. “Audience Expansion for Online Social Network Advertising”. B: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. KDD ’16. San Francisco, California, USA: Association for Computing Machinery, aвг. 2016. P. 165–174.
Yudan Liu et al. “Real-time Attention Based Look-alike Model for Recommender System”. B: (июнь 2019). arXiv: 1906.05022 [cs.IR].
Zhining Liu et al. “Two-Stage Audience Expansion for Financial Targeting in Marketing”. B: Proceedings of the 29th ACM International Conference on Information & Knowledge Management. CIKM ’20. Virtual Event, Ireland: Association for Computing Machinery, oкт. 2020. P. 2629–2636.
Qiang Ma, Musen Wen, Datong Chen. A sub-linear, massive-scale lookalike audience extension system. http://proceedings.mlr.press/v53/ma16.pdf. Accessed: 2022-2-7.
Qiang Ma et al. “Score Look-alike Audiences”. B: ().
Ashish Mangalampalli et al. “A feature-pair-based associative classification approach to look-alike modeling for conversion-oriented user-targeting in tail campaigns”. B: Proceedings of the 20th international conference companion on World wide web.WWW’11. Hyderabad, India: Association for Computing Machinery, мapт 2011. P. 85–86.
Ashish Mangalampalli et al. “A feature-pair-based associative classification approach to look-alike modeling for conversion-oriented user-targeting in tail campaigns”. B: Proceedings of the 20th international conference companion on World wide web – WWW ’11. Hyderabad, India: ACM Press, 2011.
Hongyuan Mei, Jason Eisner. “The Neural Hawkes Process: A Neurally Self-Modulating Multivariate Point Process”. B: NeurIPS. 2017.
Lin Miao, Mark Last, Marina Litvak. “Tracking social media during the COVID-19 pandemic: The case study of lockdown in New York State”. B: Expert Systems with Applications. 2022. V. 187. P. 115797. issn: 0957-4174. https://doi.org/10.1016/j.eswa.2021.115797. url: https://www.sciencedirect.com/science/article/pii/S0957417421011659.
Du Nan et al. “Recurrent Marked Temporal Point Processes: Embedding Event History to Vector”. B: KDD’16. 2016.
Aaron van den Oord et al. “WaveNet: A Generative Model for Raw Audio”. B: 9th ISCA Speech Synthesis Workshop.
Sandeep Pandey et al. “Learning to target: what works for behavioral targeting”. B: Proceedings of the 20th ACM international conference on Information and knowledge management. CIKM ’11. Glasgow, Scotland, UK: Association for Computing Machinery, oкт. 2011. P. 1805–1814.
Perlich C. et al. “Machine learning for targeted display advertising: Transfer learning in action”. B: ().
Yan Qu et al. “Systems and methods for generating expanded user segments”. 8655695. Фeвp. 2014.
Ernest Kirubakaran Selvaraj. Multigraph-Lookalike. en.
Ernest Kirubakaran Selvaraj et al. “Multigraph Approach Towards a Scalable, Robust look-alike Audience Extension System”. B. 2021.
Oleksandr Shchur et al. “Neural temporal point processes: A review”. B: arXiv preprint arXiv:2104.03528. 2021.
Jianqiang Shen, Sahin Cem Geyik, Ali Dasdan. “Effective Audience Extension in Online Advertising”. B: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. KDD ’15. Sydney, NSW, Australia: Association for Computing Machinery, aвг. 2015. P. 2099–2108.
Hui Shi et al. “Continuous CNN for nonuniform time series”. B: ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE. 2021. P. 3550–3554.
Victoria Snorovikhina, Alexey Zaytsev. “Unsupervised anomaly detection for discrete sequence healthcare data”. B: International Conference on Analysis of Images, Social Networks and Texts. Springer. 2020. P. 391–403.
Yi Tay et al. “Efficient transformers: A survey”. B: ACM Computing Surveys. 2022. V. 55.6. P. 1–28.
Yi Tay et al. “Long range arena: A benchmark for efficient transformers”. B: arXiv preprint arXiv:2011.04006. 2020.
Nikolaos Tziortziotis et al. “Audience expansion based on user browsing history”. B: 2021 International Joint Conference on Neural Networks (IJCNN). 2021. P. 1–8. https://doi.org/10.1109/IJCNN52387.2021.9533392.
Yansong Wang et al. “CasSeqGCN: Combining network structure and temporal sequence to predict information cascades”. B: Expert Systems with Applications. 2022. V. 206. P. 117693. issn: 0957-4174. https://doi.org/10.1016/j.eswa.2022.117693. url: https://www.sciencedirect.com/science/article/pii/S095741742200985X.
Hongteng Xu, Hongyuan Zha. “A Dirichlet mixture model of Hawkes processes for event sequence clustering”. B: Advances in Neural Information Processing Systems. 2017. P. 1354–1363.
Junchi Yan. “Recent advance in temporal point process: from machine learning perspective”. B: SJTU Technical Report. 2019.
Carl Yang et al. “I Know You’ll Be Back: Interpretable New User Clustering and Churn Prediction on a Mobile Social Application”. B: Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. KDD ’18. London, United Kingdom: Association for Computing Machinery, июль 2018. P. 914–922.
Wang Yaqing et al. “EANN: Event Adversarial Neural Networks for Multi-Modal Fake News Detection”. B: KDD. 2018.
James Zhang et al. “Dynamic time warp-based clustering: Application of machine learning algorithms to simulation input modelling”. B: Expert Systems with Applications. 2021. V. 186. P. 115684. issn: 0957-4174. https://doi.org/10.1016/j.eswa.2021.115684. url: https://www.sciencedirect.com/science/article/pii/S0957417421010691.
Yunhao Zhang et al. “Learning mixture of neural temporal point processes for multi-dimensional event sequence clustering”. B: Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, Vienna, Austria. 2022. P. 23–29.
Zhihua Zhang et al. “Learning a Multivariate Gaussian Mixture Model with the Reversible Jump MCMC Algorithm”. B: Statistics and Computing 14 (aпp. 2004). https://doi.org/10.1023/B:STCO.0000039484.36470.41
Yongchun Zhu et al. “Learning to Expand Audience via Meta Hybrid Experts and Critics for Recommendation and Advertising”. B: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. New York, NY, USA: Association for Computing Machinery, aвг. 2021. P. 4005–4013.
Chenyi Zhuang et al. “Hubble: An Industrial System for Audience Expansion in Mobile Marketing”. B: Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. KDD ’20. Virtual Event, CA, USA: Association for Computing Machinery, 2020. P. 2455–2463. isbn: 9781450379984. https://doi.org/10.1145/3394486.3403295. url: 10.1145/3394486.3403295
Chenyi Zhuang et al. “Hubble: An Industrial System for Audience Expansion in Mobile Marketing”. B: Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York, NY, USA: Association for Computing Machinery, aвг. 2020. P. 2455–2463.
Jinfeng Zhuang et al. “PinText 2: Attentive bag of annotations embedding”. B: ().
Simiao Zuo et al. “Transformer Hawkes process”. B: International conference on machine learning. PMLR. 2020. P. 11692–11702.
Дополнительные материалы отсутствуют.
Инструменты
Доклады Российской академии наук. Математика, информатика, процессы управления