Whole Page Optimization with Global Constraints

Ding, Weicong; Govindaraj, Dinesh; Vishwanathan, S. V. N.

doi:10.1145/3292500.3330675

Cited by 23 publications

(20 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Earlier studies [10,27] demonstrate that user engagement increases when users are presented with a diverse selection of content. In general, page-level optimization needs to satisfy multiple important goals [13,17,21,26]. First, we must recommend relevant modules to users based on their previous shopping history.…”

Section: Slotmentioning

confidence: 99%

“…However, this requires evaluating a combinatorially large number of page presentations, which quickly becomes prohibitively expensive, especially due to page loading latency considerations when a large set of modules are involved. Recent works approximate the combinatorial space by considering pairwise interactions among modules [13,17,26], and explore simple greedy procedures to learn the inter-dependencies [17]. However, these models are not scalable when the given contextual features have high cardinality (e.g.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Page-level Optimization of e-Commerce Item Recommendations

Xin

et al. 2021

Fifteenth ACM Conference on Recommender Systems

View full text Add to dashboard Cite

The item details page (IDP) is a web page on an e-commerce website that provides information on a specific product or item listing. Just below the details of the item on this page, the buyer can usually find recommendations for other relevant items. These are typically in the form of a series of modules or carousels, with each module containing a set of recommended items. The selection and ordering of these item recommendation modules are intended to increase discover-ability of relevant items and encourage greater user engagement, while simultaneously showcasing diversity of inventory and satisfying other business objectives. Item recommendation modules on the IDP are often curated and statically configured for all customers, ignoring opportunities for personalization. In this paper, we present a scalable end-to-end production system to optimize the personalized selection and ordering of item recommendation modules on the IDP in real-time by utilizing deep neural networks. Through extensive offline experimentation and online A/B testing, we show that our proposed system achieves significantly higher click-through and conversion rates compared to other existing methods. In our online A/B test, our framework improved click-through rate by 2.48% and purchase-through rate by 7.34% over a static configuration. CCS Concepts: • Information systems → Learning to rank; Novelty in information retrieval; • Computing methodologies → Ranking; • Applied computing → E-commerce infrastructure.

show abstract

Section: Slotmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Page-level Optimization of e-Commerce Item Recommendations

Xin

et al. 2021

Fifteenth ACM Conference on Recommender Systems

View full text Add to dashboard Cite

show abstract

“…In particular, they design a real-time model to understand the user preferences and adjust the carousels displayed. Ding et al [5] target the problem of optimizing the carousel homepage of Amazon Prime Video, a video streaming service. They point out that business constraints may fix carousels in certain positions of the grid, so the evaluation should take into account how the other carousel intersect with the fixed ones and how they improve the user experience.…”

Section: Related Workmentioning

confidence: 99%

“…In fact, in video-ondemand services (e.g., Netflix, Amazon Prime Video) and music streaming platforms (e.g., Spotify) users are frequently provided with several recommendation lists or carousels, each with a certain theme e.g., recently added, originals, trending, editorially curated (see Figure 1). In this scenario, the user satisfaction depends on all the carousels that are shown, not just one, and there is significant industrial interest in finding effective strategies to select which carousels to display [2,5,15]. In order to represent such a scenario more closely, it is necessary to take into account how the recommendations in the various carousels complement each other and how the user explore a two-dimensional interface.…”

Section: Introductionmentioning

confidence: 99%

Measuring the User Satisfaction in a Recommendation Interface with Multiple Carousels

Felicioni

Dacrema

Cremonesi

2021

ACM International Conference on Interactive Media Experiences

View full text Add to dashboard Cite

It is common for video-on-demand and music streaming services to adopt a user interface composed of several recommendation lists, i.e., widgets or swipeable carousels, each generated according to a specific criterion or algorithm (e.g., most recent, top popular, recommended for you, editors' choice, etc.). Selecting the appropriate combination of carousel has significant impact on user satisfaction. A crucial aspect of this user interface is that to measure the relevance a new carousel for the user it is not sufficient to account solely for its individual quality. Instead, it should be considered that other carousels will already be present in the interface. This is not considered by traditional evaluation protocols for recommenders systems, in which each carousel is evaluated in isolation, regardless of (i) which other carousels are displayed to the user and (ii) the relative position of the carousel with respect to other carousels. Hence, we propose a two-dimensional evaluation protocol for a carousel setting that will measure the quality of a recommendation carousel based on how much it improves upon the quality of an already available set of carousels. Our evaluation protocol takes into account also the position bias, i.e., users do not explore the carousels sequentially, but rather concentrate on the top-left corner of the screen.We report experiments on the movie domain and notice that under a carousel setting the definition of which criteria has to be preferred to generate a list of recommended items changes with respect to what is commonly understood.

show abstract

“…given a list of recommendations). Given item-wise values for a certain metric of interest, they are then able to enforce constraints which require that the expected value of the metric (under a specific model of position-dependent attention or exposure probabilities) is larger than a threshold [1,10,19] -these have been termed exposure constraints in previous work [19]. This has a distinct advantage of being able to guarantee that constraints are satisfied on a per-request basis, i.e.…”

Section: Related Workmentioning

confidence: 99%

The NodeHopper: Enabling Low Latency Ranking with Constraints via a Fast Dual Solver

Zhernov

Dvijotham

Lobov

et al. 2020

Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining

View full text Add to dashboard Cite

Modern recommender systems need to deal with multiple objectives like balancing user engagement with recommending diverse and fresh content. An appealing way to optimally trade these off is by imposing constraints on the ranking according to which items are presented to a user. This results in a constrained ranking optimization problem that can be solved as a linear program (LP). However, off-the-shelf LP solvers are unable to meet the severe latency constraints in systems that serve live traffic. To address this challenge, we exploit the structure of the dual optimization problem to develop a fast solver. We analyze theoretical properties of our solver and show experimentally that it is able to solve constrained ranking problems on synthetic and real-world recommendation datasets an order of magnitude faster than off-the-shelf solvers, thereby enabling their deployment under severe latency constraints.

show abstract

Whole Page Optimization with Global Constraints

Cited by 23 publications

References 21 publications

Page-level Optimization of e-Commerce Item Recommendations

Page-level Optimization of e-Commerce Item Recommendations

Measuring the User Satisfaction in a Recommendation Interface with Multiple Carousels

The NodeHopper: Enabling Low Latency Ranking with Constraints via a Fast Dual Solver

Contact Info

Product

Resources

About