Authors: Sejoon Oh, Moumita Bhattacharya, Yesu Feng, Sudarshan Lamkhede, Ko-Jen Hsiao, and Justin Basilico
Recommender programs have develop into important parts of digital companies throughout e-commerce, streaming media, and social networks [1, 2]. At Netflix, these programs drive important product and enterprise influence by connecting members with related content material on the proper time [3, 4]. Whereas our advice basis mannequin (FM) has made substantial progress in understanding consumer preferences via large-scale studying from interplay histories (please confer with this article about FM @ Netflix), there is a chance to additional improve its capabilities. By extending FM to include the prediction of underlying consumer intents, we purpose to complement its understanding of consumer classes past next-item prediction, thereby providing a extra complete and nuanced advice expertise.
Latest analysis has highlighted the significance of understanding consumer intent in on-line platforms [5, 6, 7, 8]. As Xia et al. [8] demonstrated at Pinterest, predicting a consumer’s future intent can result in extra correct and personalised suggestions. Nevertheless, present intent prediction approaches sometimes make use of easy multi-task studying that provides intent prediction heads to next-item prediction fashions with out establishing a hierarchical relationship between these duties.
To handle these limitations, we introduce FM-Intent, a novel advice mannequin that enhances our basis mannequin via hierarchical multi-task studying. FM-Intent captures a consumer’s latent session intent utilizing each short-term and long-term implicit indicators as proxies, then leverages this intent prediction to enhance next-item suggestions. Not like typical approaches, FM-Intent establishes a transparent hierarchy the place intent predictions instantly inform merchandise suggestions, making a extra coherent and efficient advice pipeline.
FM-Intent makes three key contributions:
- A novel advice mannequin that captures consumer intent on the Netflix platform and enhances next-item prediction utilizing this intent info.
- A hierarchical multi-task studying strategy that successfully fashions each short-term and long-term consumer pursuits.
- Complete experimental validation displaying important efficiency enhancements over state-of-the-art fashions, together with our basis mannequin.
Within the Netflix ecosystem, consumer intent manifests via varied interplay metadata, as illustrated in Determine 1. FM-Intent leverages these implicit indicators to foretell each consumer intent and next-item suggestions.
Determine 1: Overview of consumer engagement knowledge in Netflix. Person intent might be related to a number of interplay metadata. We leverage varied implicit indicators to foretell consumer intent and next-item.
In Netflix, there might be a number of varieties of consumer intents. For example,
Motion Sort: Classes reflecting what customers intend to do on Netflix, reminiscent of discovering new content material versus persevering with beforehand began content material. For instance, when a member performs a follow-up episode of one thing they had been already watching, this may be categorized as “proceed watching” intent.
Style Choice: The pre-defined style labels (e.g., Motion, Thriller, Comedy) that point out a consumer’s content material preferences throughout a session. These preferences can shift considerably between classes, even for a similar consumer.
Film/Present Sort: Whether or not a consumer is in search of a film (sometimes a single, longer viewing expertise) or a TV present (probably a number of episodes of shorter length).
Time-since-release: Whether or not the consumer prefers newly launched content material, latest content material (e.g., between every week and a month), or evergreen catalog titles.
These dimensions function proxies for the latent consumer intent, which is usually indirectly observable however essential for offering related suggestions.
FM-Intent employs a hierarchical multi-task studying strategy with three main parts, as illustrated in Determine 2.
Determine 2: An architectural illustration of our hierarchical multi-task studying mannequin FM-Intent for consumer intent and merchandise predictions. We use ground-truth intent and item-ID labels to optimize predictions.
1. Enter Function Sequence Formation
The primary part constructs wealthy enter options by combining interplay metadata. The enter function for every interplay combines categorical embeddings and numerical options, making a complete illustration of consumer habits.
2. Person Intent Prediction
The intent prediction part processes the enter function sequence via a Transformer encoder and generates predictions for a number of intent indicators.
The Transformer encoder successfully fashions the long-term curiosity of customers via multi-head consideration mechanisms. For every prediction activity, the intent encoding is reworked into prediction scores through fully-connected layers.
A key innovation in FM-Intent is the attention-based aggregation of particular person intent predictions. This strategy generates a complete intent embedding that captures the relative significance of various intent indicators for every consumer, offering worthwhile insights for personalization and rationalization.
3. Subsequent-Merchandise Prediction with Hierarchical Multi-Job Studying
The ultimate part combines the enter options with the consumer intent embedding to make extra correct next-item suggestions.
FM-Intent employs hierarchical multi-task studying the place intent predictions are performed first, and their outcomes are used as enter options for the next-item prediction activity. This hierarchical relationship ensures that the next-item suggestions are knowledgeable by the anticipated consumer intent, making a extra coherent and efficient advice mannequin.
We performed complete offline experiments on sampled Netflix consumer engagement knowledge to judge FM-Intent’s efficiency. Observe that FM-Intent makes use of a a lot smaller dataset for coaching in comparison with the FM manufacturing mannequin resulting from its complicated hierarchical prediction structure.
Subsequent-Merchandise and Subsequent-Intent Prediction Accuracy
Desk 1 compares FM-Intent with a number of state-of-the-art sequential advice fashions, together with our manufacturing mannequin (FM-Intent-V0).
Desk 1: Subsequent-item and next-intent prediction outcomes of baselines and our proposed technique FM-Intent on the Netflix consumer engagement dataset.
All metrics are represented as relative % enhancements in comparison with the SOTA baseline: TransAct. N/A signifies {that a} mannequin isn’t able to predicting a sure intent. Observe that we added further fully-connected layers to LSTM, GRU, and Transformer baselines with a view to predict consumer intent, whereas we used unique implementations for different baselines. FM-Intent demonstrates statistically important enchancment of seven.4% in next-item prediction accuracy in comparison with the very best baseline (TransAct).
Most baseline fashions present restricted efficiency as they both can not predict consumer intent or can not incorporate intent predictions into next-item suggestions. Our manufacturing mannequin (FM-Intent-V0) performs nicely however lacks the flexibility to foretell and leverage consumer intent. Observe that FM-Intent-V0 is skilled with a smaller dataset for a good comparability with different fashions; the precise manufacturing mannequin is skilled with a a lot bigger dataset.
Determine 3: Okay-means++ (Okay=10) clustering of consumer intent embeddings discovered by FM-Intent; FM-Intent finds distinctive clusters of customers that share the same intent.
FM-Intent generates significant consumer intent embeddings that can be utilized for clustering customers with related intents. Determine 3 visualizes 10 distinct clusters recognized via Okay-means++ clustering. These clusters reveal significant consumer segments with distinct viewing patterns:
- Customers who primarily uncover new content material versus those that proceed watching latest/favourite content material.
- Style lovers (e.g., anime/children content material viewers).
- Customers with particular viewing patterns (e.g., Rewatchers versus informal viewers).
FM-Intent has been efficiently built-in into Netflix’s advice ecosystem, might be leveraged for a number of downstream purposes:
Customized UI Optimization: The expected consumer intent may inform the format and content material choice on the Netflix homepage, emphasizing completely different rows primarily based on whether or not customers are in discovery mode, continue-watching mode, or exploring particular genres.
Analytics and Person Understanding: Intent embeddings and clusters present worthwhile insights into viewing patterns and preferences, informing content material acquisition and manufacturing selections.
Enhanced Advice Alerts: Intent predictions function options for different advice fashions, enhancing their accuracy and relevance.
Search Optimization: Actual-time intent predictions assist prioritize search outcomes primarily based on the consumer’s present session intent.
FM-Intent represents an development in Netflix’s advice capabilities by enhancing them with hierarchical multi-task studying for consumer intent prediction. Our complete experiments reveal that FM-Intent considerably outperforms state-of-the-art fashions, together with our prior basis mannequin that centered solely on next-item prediction. By understanding not simply what customers would possibly watch subsequent however what underlying intents customers have, we are able to present extra personalised, related, and satisfying suggestions.
We thank our beautiful colleagues within the Basis Mannequin crew & AIMS org. for his or her worthwhile suggestions and discussions. We additionally thank our associate groups for getting this up and operating in manufacturing.
[1] Amatriain, X., & Basilico, J. (2015). Recommender programs in business: A netflix case examine. In Recommender programs handbook (pp. 385–419). Springer.
[2] Gomez-Uribe, C. A., & Hunt, N. (2015). The netflix recommender system: Algorithms, enterprise worth, and innovation. ACM Transactions on Administration Data Programs (TMIS), 6(4), 1–19.
[3] Jannach, D., & Jugovac, M. (2019). Measuring the enterprise worth of recommender programs. ACM Transactions on Administration Data Programs (TMIS), 10(4), 1–23.
[4] Bhattacharya, M., & Lamkhede, S. (2022). Augmenting Netflix Search with In-Session Tailored Suggestions. In Proceedings of the sixteenth ACM Convention on Recommender Programs (pp. 542–545).
[5] Chen, Y., Liu, Z., Li, J., McAuley, J., & Xiong, C. (2022). Intent contrastive studying for sequential advice. In Proceedings of the ACM Internet Convention 2022 (pp. 2172–2182).
[6] Ding, Y., Ma, Y., Wong, W. Okay., & Chua, T. S. (2021). Modeling on the spot consumer intent and content-level transition for sequential trend advice. IEEE Transactions on Multimedia, 24, 2687–2700.
[7] Liu, Z., Chen, H., Solar, F., Xie, X., Gao, J., Ding, B., & Shen, Y. (2021). Intent choice decoupling for consumer illustration on on-line recommender system. In Proceedings of the Twenty-Ninth Worldwide Convention on Worldwide Joint Conferences on Synthetic Intelligence (pp. 2575–2582).
[8] Xia, X., Eksombatchai, P., Pancha, N., Badani, D. D., Wang, P. W., Gu, N., Joshi, S. V., Farahpour, N., Zhang, Z., & Zhai, A. (2023). TransAct: Transformer-based Realtime Person Motion Mannequin for Advice at Pinterest. In Proceedings of the twenty ninth ACM SIGKDD Convention on Information Discovery and Knowledge Mining (pp. 5249–5259).
