In today's fast-paced world, customers increasingly value quick and reliable delivery services, with many prioritizing speed as a decisive factor in their purchasing decisions. E-commerce stores serve customers through specialized programs ensuring delivery within same day. Facilitated by strategically placed delivery networks, this provides an ultra-fast delivery experience to the end customers enabling them to receive their orders within the same day in a chosen fixed size window. While intra-day deliveries conclusively improve the customer experience, in the age of quick commerce, customers want the orders faster, and many scenarios (for e.g. periods of the day when customers would be traveling to school or work) make long static windows inconvenient for customers. However, faster deliveries increase the cost of shipping. In this work, we leverage the observation that in many instances orders reach ahead of time because of the geographical proximity of the shipping address and the order density in the neighborhood. This presents an opportunity to improve the delivery experience of customers without incurring any additional costs for customer or the seller. We present a framework to recommend dynamic, faster delivery time slots to customers. We create multiple heterogeneous views of order-to-delivery data capturing the spatial and spatio-temporal aspects of the data, and leverage a novel deep view interaction network which computes the higher order interactions among the views. The proposed model outperforms multiple representative baselines and allows us to predict narrower slots for 60%+ eligible orders for the locale under experimentation. During a 21 day online A/B test, the treatment recorded a significant gain of +17 bps in units, +21 bps for views and +19 bps increase in search interactions, establishing the efficacy of the framework.

In online businesses, personalization of site content is crucial for providing a better user experience and increasing customer engagement. Machine learning algorithms are often used to analyze customer data such as browsing behavior, purchase history to tailor the website content to each individual customer's preferences and needs. However, measuring the success of these personalized experiences can be challenging. While the ultimate goal is to convert customer visits into purchase sessions, tracking individual customer behavior can be difficult at scale. As a result, businesses often rely on aggregate metrics such as site-wide conversion rates, sales, and revenue to evaluate the effectiveness of their personalization efforts. However, it's important to understand individual customers' experiences with these ML-optimized pages. To address this, we propose a supervised ML model that quantifies customer engagement while browsing auto-optimized web pages, by building a customer site interaction score (CSI score). We first introduce a novel representation of customer click logs as a tree data structure induced by the webpage's DOM structure. Then, we propose a novel attention model on the tree structure that performs vertical attention across the depth of the tree and horizontal attention across the sequence of trees to summarize customer interactions. The effectiveness of the proposed approach is evaluated using click logs obtained from the e-commerce domain.

Internet users actively search for trending products on various social media services like Instagram and YouTube which serve as popular hubs for discovering and exploring fashionable and popular items. It is imperative for e-commerce giants to have the capability to accurately identify, predict and subsequently showcase these trending products to the customers. E-commerce stores can effectively cater to the evolving demands of the customer base and enhance the overall shopping experience by offering recent and most sought-after products in a timely manner. In this work we propose a framework for predicting and surfacing trending products in e-commerce stores, the first of its kind to the best of our knowledge. We begin by defining what constitutes a trending product using sound statistical tests. We then introduce a machine learning-based early trend prediction system called TrendSpotter to help users identify upcoming product trends. TrendSpotter is a unique adaptation of the state-of-the-art InceptionTime model\citeInceptionTime that predicts the future popularity of a product based on its current customer engagement, such as clicks, purchases, and other relevant product attributes. The effectiveness of our approach is demonstrated through A/B tests, where we first showcase the effectiveness of our statistical test based labeling strategy, resulting in an incremental sales lift of 59 bps\footnotebps or basis points are a measure of percentages. 1 bps = 0.01% across two experiments on home page and search page. Subsequently, we conduct a comparison between our machine learning model and the statistical labeling baseline and observe an additional sales gain of 14 bps, reflecting the importance of early identification of trending products.

Change Point Detection (CPD) models are used to identify abrupt changes in the distribution of a data stream and have a widespread practical use. CPD methods generally compare the distribution of data sequences before and after a given time step to infer if there is a shift in distribution at the said time step. Numerous divergence measures, which measure distance between data distributions of sequence pairs, have been proposed for CPD \citeMStatisticNIPS, BergCPD and often the choice of divergence measure depends on the data used. Density Ratio Estimation (DRE) \citeRelDivCPD,BergCPD can be used to estimate a broad family of f-divergences, which includes widely used CPD divergences like Kullback-Leibler (KL) and Pearson, and thus DRE is a popular approach for CPD. In this work, we improve upon the existing DRE techniques for CPD, by proposing a novel objective that combines DRE seamlessly with adversarial sample generation. The adversarial samples allows for a robust CPD with DRE to track subtle changes in distribution, leading to a reduction in false negatives. We experiment on a wide variety of real-world, public benchmark datasets to show that our approach improves upon existing state-of-the-art (SoTA) methods, including DRE based CPD methods, by demonstrating an \sim 5% increase in F-score.

Finding the right product on e-commerce websites with millions of products is a daunting task for a large set of customers. On the search page, product attribute filters a.k.a. "refinements" emerge as a convenient navigational option for customers to narrow down the search results along product attributes of their choice (e.g., Material:Cotton, Color:Black for 'shirt'). However, on mobile devices, refinements are not easily discoverable due to lack of screen space. To improve discoverability, contextually relevant refinements are suggested in-line on search page by refinement recommendation systems. Existing works on refinement recommendations primarily rely on the search context as input, and are trained using aggregated refinement preferences 'explicitly' expressed by customers. These solutions fail to capture 'implicit' preferences expressed during the customer shopping mission through in-session browsing activity. In this paper, we propose a session-based recommendation system (SBRS) which recommends refinements by inferring product attribute preferences of customers based on the sequence of products viewed earlier in the session. For the task of refinement recommendation, we propose a) AttriBERT, a model which extends BERT architecture to learn from the attribute values of products and b) a novel product representation strategy, which represents each product as a dictionary of attribute:value pairs (e.g., RAM Size:64GB). We evaluate our approach on RecSys 2022 Challenge and Amazon e-commerce datasets. Our approach consistently outperforms various state-of-the-art sequence models on the task of session-based refinement recommendation.

Numerous problems of practical significance such as clickthrough rate (CTR) prediction, forecasting, tagging and so on, involve complex interaction of various user, item and context features. Manual feature engineering has been used in the past to model these combinatorial features but it requires domain expertise and becomes prohibitively expensive as the number of features increases. Feedforward neural networks alleviate the need for manual feature engineering to a large extent and have shown impressive performance across multiple domains due to their ability to learn arbitrary functions. Despite multiple layers of non-linear projections, neural networks are limited in their ability to efficiently model functions with higher order interaction terms. In recent years, Factorization Machines and its variants have been proposed to explicitly capture higher order combinatorial interactions. However not all feature interactions are equally important, and in sparse data settings, without a suitable suppression mechanism, this might result into noisy terms during inference and hurt model generalization. In this work we present Dual Attentional Higher Order Factorization Machine (DA-HoFM), a unified attentional higher order factorization machine which leverages a compositional architecture to compute higher order terms with complexity linear in terms of maximum interaction degree. Equipped with sparse dual attention mechanism, DA-HoFM summarizes interaction terms at each layer, and is able to efficiently select important higher order terms. We empirically demonstrate effectiveness of our proposed models on the task of CTR prediction, where our model exhibits superior performance compared to the recent state-of-the-art models, outperforming them by up to 6.7% on the logloss metric.

Estimating the relevance of documents based on the user feedback is an essential component of search, retrieval and ranking problems. User click modeling in search has focused primarily on factoring out the position bias. It is easy to see that the query type (generic queries vs specific queries) and user intent (purchase vs exploration) also introduce a bias in the click signal. In other words, the results not matching with the user intent will not be clicked. In this paper, we outline a technique to model the interplay of query, user intent and position bias with respect to the relevance of the retrieved search results. In particular, we define two intents namely purchase and explore, and estimate the relevance of the documents with respect to these two intents. We also relate them to the relevance estimates from considering only the position bias. We empirically demonstrate the effectiveness of the proposed approach by comparing its performance against the well-known CoEC measure and the recently proposed factor model approach for relevance estimation.

Community detection has emerged as an attractive topic due to the increasing need to understand and manage the networked data of tremendous magnitude. Networked data usually consists of links between the entities and the attributes for describing the entities. Various approaches have been proposed for detecting communities by utilizing the link information and/or attribute information. In this work, we study the problem of community detection for networked data with additional authorship information. By authorship, each entity in the network is authored by another type of entities (e.g., wiki pages are edited by users, products are purchased by customers), to which we refer as authors. Communities of entities are affected by their authors, e.g., two entities that are associated with the same author tend to belong to the same community. Therefore leveraging the authorship information would help us better detect the communities in the networked data. However, it also brings new challenges to community detection. The foremost question is how to model the correlation between communities and authorships. In this work, we address this question by proposing probabilistic models based on the popularity link model [1], which is demonstrated to yield encouraging results for community detection. We employ two methods for modeling the authorships: (i) the first one generates the authorships independently from links by community memberships and popularities of authors by analogy of the popularity link model; (ii) the second one models the links between entities based on authorships together with community memberships and popularities of nodes, which is an analog of previous author-topic model. Upon the basic models, we explore several extensions including (i) we model the community memberships of authors by that of their authored entities to reduce the number of redundant parameters; and (ii) we model the communities memberships of entities and/or authors by their attributes using a discriminative approach. We demonstrate the effectiveness of the proposed models by empirical studies.

Real-world data are often riddled with data quality problems such as noise, outliers and missing values, which present significant challenges for supervised learning algorithms to effectively classify them. This paper explores the ill-effects of inapplicable features on the performance of supervised learning algorithms. In particular, we highlight the difference between missing and inapplicable feature values. We argue that the current approaches for dealing with missing values, which are mostly based on single or multiple imputation methods, are insufficient to handle inapplicable features, especially those that are continuous valued. We also illustrate how current tree-based and kernel-based classifiers can be adversely affected by the presence of such features if not handled appropriately. Finally, we propose methods to extend existing tree-based and kernel-based classifiers to deal with the inapplicable continuous-valued features.

Malware detection from network traffic flows is a challenging problem due to data irregularity issues such as imbalanced class distribution, noise, missing values, and heterogeneous types of features. To address these challenges, this paper presents a two-stage classification approach for malware detection. The framework initially employs random forest as a macro-level classifier to separate the malicious from non-malicious network flows, followed by a collection of one-class support vector machine classifiers to identify the specific type of malware. A novel tree-based feature construction approach is proposed to deal with data imperfection issues. As the performance of the support vector machine classifier often depends on the kernel function used to compute the similarity between every pair of data points, designing an appropriate kernel is essential for accurate identification of malware classes. We present a simple algorithm to construct a weighted linear kernel on the tree transformed features and demonstrate its effectiveness in detecting malware from real network traffic data.

Link prediction is a challenging task due to the inherent skew ness of network data. Typical link prediction methods can be categorized as either local or global. Local methods consider the link structure in the immediate neighborhood of a node pair to determine the presence or absence of a link, whereas global methods utilize information from the whole network. This paper presents a community (cluster) level link prediction method without the need to explicitly identify the communities in a network. Specifically, a variable-cost loss function is defined to address the data skew ness problem. We provide theoretical proof that shows the equivalence between maximizing the well-known modularity measure used in community detection and minimizing a special case of the proposed loss function. As a result, any link prediction method designed to optimize the loss function would result in more links being predicted within a community than between communities. We design a boosting algorithm to minimize the loss function and present an approach to scale-up the algorithm by decomposing the network into smaller partitions and aggregating the weak learners constructed from each partition. Experimental results show that our proposed Link Boost algorithm consistently performs as good as or better than many existing methods when evaluated on 4 real-world network datasets.

With the rapid proliferation of online social networks, the need for newer class of learning algorithm to simultaneously deal with multiple related networks has become increasingly important. This paper proposes an approach for multi-task learning in multiple related networks, where in we perform different tasks such as classification on one network and clustering on the other. We show that the framework can be extended to incorporate prior information about the correspondences between the clusters and classes in different networks. We have performed experiments on real-world data sets to demonstrate the effectiveness of the proposed framework.

Identifying cohesive subgroups in networks, also known as clustering is an active area of research in link mining with many practical applications. However, most of the early work in this area has focused on partitioning a single network or a bipartite graph into clusters/communities. This paper presents a framework that simultaneously clusters nodes from multiple related networks and learns the correspondences between subgroups in different networks. The framework also allows the incorporation of prior information about potential relationships between the subgroups. We have performed extensive experiments on both synthetic and real-life data sets to evaluate the effectiveness of our framework. Our results show superior performance of simultaneous clustering over independent clustering of individual networks.