In this paper, we describe a solution to tackle a common set of challenges in e-commerce, which arise from the fact that new products are continually being added to the catalogue. The challenges involve properly personalising the customer experience, forecasting demand and planning the product range. We argue that the foundational piece to solve all of these problems is having consistent and detailed information about each product, information that is rarely available or consistent given the multitude of suppliers and types of products. We describe in detail the architecture and methodology implemented at ASOS, one of the world's largest fashion e-commerce retailers, to tackle this problem. We then show how this quantitative understanding of the products can be leveraged to improve recommendations in a hybrid recommender system approach.
In many personalized recommendation problems available data consists only of positive interactions (implicit feedback) between users and items. This problem is also known as One-Class Collaborative Filtering (OC-CF). Linear models usually achieve state-of-the-art performances on OC-CF problems and many efforts have been devoted to build more expressive and complex representations able to improve the recommendations. Recent analysis show that collaborative filtering (CF) datasets have peculiar characteristics such as high sparsity and a long tailed distribution of the ratings. In this paper we propose a boolean kernel, called Disjunctive kernel, which is less expressive than the linear one but it is able to alleviate the sparsity issue in CF contexts. The embedding of this kernel is composed by all the combinations of a certain arity d of the input variables, and these combined features are semantically interpreted as disjunctions of the input variables. Experiments on several CF datasets show the effectiveness and the efficiency of the proposed kernel.
Recommendation systems are unique as they show a conflation of compute and memory intensity due to their deep learning and massive embedding tables. Training these models typically involve a hybrid CPU-GPU mode, where GPUs accelerate the deep learning portion and the CPUs store and process the memory-intensive embedding tables. The hybrid mode incurs a substantial CPU-to-GPU transfer time and relies on main memory bandwidth to feed embeddings to GPU for deep learning acceleration. Alternatively, we can store the entire embeddings across GPUs to avoid the transfer time and utilize the GPU's High Bandwidth Memory (HBM). This approach requires GPU-to-GPU backend communication and scales the number of GPUs with the size of the embedding tables. To overcome these concerns, this paper offers a heterogeneous acceleration pipeline, called Hotline. Hotline leverages the insight that only a small number of embedding entries are accessed frequently, and can easily fit in a single GPU's HBM. Hotline implements a data-aware and model-aware scheduling pipeline that utilizes the (1) CPU main memory for not-frequently-accessed embeddings and (2) GPUs' local memory for frequently-accessed embeddings. Hotline improves the training throughput by dynamically stitching the execution of popular and not-popular inputs through a novel hardware accelerator and feeding to the GPUs. Results on real-world datasets and recommender models show that Hotline reduces the average training time by 3x and 1.8x in comparison to Intel-optimized CPU-GPU DLRM and HugeCTR-optimized GPU-only baseline, respectively. Hotline increases the overall training throughput to 35.7 epochs/hour in comparison to 5.3 epochs/hour for the Intel-optimized DLRM baseline
Maternal and child mortality is a public health problem that disproportionately affects low- and middle-income countries. Every day, 800 women and 6,700 newborns die from complications related to pregnancy or childbirth. And for every maternal death, about 20 women suffer serious birth injuries. However, nearly all of these deaths and negative health outcomes are preventable. Midwives are key to revert this situation, and thus it is essential to strengthen their capacities and the quality of their education. This is the aim of the Safe Delivery App, a digital job aid and learning tool to enhance the knowledge, confidence and skills of health practitioners. Here, we use the behavioral logs of the App to implement a recommendation system that presents each midwife with suitable contents to continue gaining expertise. We focus on predicting the click-through rate, the probability that a given user will click on a recommended content. We evaluate four deep learning models and show that all of them produce highly accurate predictions.
As machine learning becomes more widely adopted across domains, it is critical that researchers and ML engineers think about the inherent biases in the data that may be perpetuated by the model. Recently, many studies have shown that such biases are also imbibed in Graph Neural Network (GNN) models if the input graph is biased. In this work, we aim to mitigate the bias learned by GNNs through modifying the input graph. To that end, we propose FairMod, a Fair Graph Modification methodology with three formulations: the Global Fairness Optimization (GFO), Community Fairness Optimization (CFO), and Fair Edge Weighting (FEW) models. Our proposed models perform either microscopic or macroscopic edits to the input graph while training GNNs and learn node embeddings that are both accurate and fair under the context of link recommendations. We demonstrate the effectiveness of our approach on four real world datasets and show that we can improve the recommendation fairness by several factors at negligible cost to link prediction accuracy.
Recently, Recurrent Neural Networks (RNNs) have been applied to the task of session-based recommendation. These approaches use RNNs to predict the next item in a user session based on the previ- ously visited items. While some approaches consider additional item properties, we argue that item dwell time can be used as an implicit measure of user interest to improve session-based item recommen- dations. We propose an extension to existing RNN approaches that captures user dwell time in addition to the visited items and show that recommendation performance can be improved. Additionally, we investigate the usefulness of a single validation split for model selection in the case of minor improvements and find that in our case the best model is not selected and a fold-like study with different validation sets is necessary to ensure the selection of the best model.
Huge volumes of patient queries are daily generated on online health forums, rendering manual doctor allocation a labor-intensive task. To better help patients, this paper studies a novel task of doctor recommendation to enable automatic pairing of a patient to a doctor with relevant expertise. While most prior work in recommendation focuses on modeling target users from their past behavior, we can only rely on the limited words in a query to infer a patient's needs for privacy reasons. For doctor modeling, we study the joint effects of their profiles and previous dialogues with other patients and explore their interactions via self-learning. The learned doctor embeddings are further employed to estimate their capabilities of handling a patient query with a multi-head attention mechanism. For experiments, a large-scale dataset is collected from Chunyu Yisheng, a Chinese online health forum, where our model exhibits the state-of-the-art results, outperforming baselines only consider profiles and past dialogues to characterize a doctor.
Prior work on personalized recommendations has focused on exploiting explicit signals from user-specific queries, clicks, likes, and ratings. This paper investigates tapping into a different source of implicit signals of interests and tastes: online chats between users. The paper develops an expressive model and effective methods for personalizing search-based entity recommendations. User models derived from chats augment different methods for re-ranking entity answers for medium-grained queries. The paper presents specific techniques to enhance the user models by capturing domain-specific vocabularies and by entity-based expansion. Experiments are based on a collection of online chats from a controlled user study covering three domains: books, travel, food. We evaluate different configurations and compare chat-based user models against concise user profiles from questionnaires. Overall, these two variants perform on par in terms of [email protected], but each has advantages in certain domains.