Buyer intent in e-commerce is multi-faceted and is expressed through explicit attributes—such as brand, size, color, and material, rather than through general topical relevance. However, many state-of-the-art scalable query-product matching systems rely on aggregate representations, scoring a single query embedding against a single item embedding. While efficient, this aggregation frequently fails to satisfy individual attribute intent: items can be semantically related, yet violate key aspects specified in the query. In contrast, fine-grained interaction methods can better capture aspect-level constraints, but are typically too expensive due to increased run-time computation and storage costs. We propose an aspect-aware ranking framework that retrieves and resolves aspects in queries and performs fine-grained semantic affinity match against aspects in products to compute an aggregate query-product level aspect affinity score. The proposed approach integrates (i) query aspect resolution (canonicalization) using structured aspect data, (ii) a model to learn granular aspect affinity signal capturing individual aspect-level understanding; and iii) an efficient design for online serving, significantly cutting cost associated with inference speed and storage. This design preserves the scalability of two-tower retrieval while substantially improving explicit intent satisfaction.

In e-commerce shopping, aligning search results with a buyer's immediate needs and preferences presents a significant challenge, particularly in adapting search results throughout the buyer's shopping journey as they move from the initial stages of browsing to making a purchase decision or shift from one intent to another. This study presents a systematic approach to adapting e-commerce search results based on the current context. We start with basic methods and incrementally incorporate more contextual information and state-of-the-art techniques to improve the search outcomes. By applying this evolving contextual framework to items displayed on the search engine results page (SERP), we progressively align search outcomes more closely with the buyer's interests and current search intentions. Our findings demonstrate that this incremental enhancement, from simple heuristic autoregressive features to advanced sequence models, significantly improves ranker performance. The integration of contextual techniques enhances the performance of our production ranker, leading to improved search results in both offline and online A/B testing in terms of Mean Reciprocal Rank (MRR). Overall, the paper details iterative methodologies and their substantial contributions to search result contextualization on e-commerce platforms.

Public school boundaries are redrawn from time to time to ensure effective functioning of school systems. This process, also called school redistricting, is non-trivial due to (1) the presence of multiple design criteria such as capacity utilization, proximity and travel time which are hard for planners to consider simultaneously, (2) the fixed locations of schools with widely differing capacities that need to be balanced, (3) the spatial nature of the data and the need to preserve contiguity in school zones, and (4) the difficulty in quantifying local factors that may arise. Motivated by these challenges and the intricacy of the process, we propose a geospatial clustering algorithm called GeoKmeans for assisting planners in designing school boundaries such that students are assigned to proximal schools while ensuring effective utilization of school capacities. The algorithm operates on polygonal geometries and connects them into geographically contiguous school boundaries while balancing problem-specific constraints. We evaluate our approach on real-world data of two rapidly growing school districts in the US. Results indicate the efficacy of our approach in designing boundaries. Additionally, a case study is included to demonstrate the potential of GeoKmeans to assist planners in drawing boundaries.

The penetration of social media has had deep and far-reaching consequences in information production and consumption. Widespread use of social media platforms has engendered malicious users and attention seekers to spread rumors and fake news. This trend is particularly evident in various microblogging platforms where news becomes viral in a matter of hours and can lead to mass panic and confusion. One intriguing fact regarding rumors and fake news is that very often rumor stories prompt users to adopt different stances about the rumor posts. Understanding user stances in rumor posts is thus very important to identify the veracity of the underlying content. While rumor veracity and stance detection have been viewed as disjoint tasks we demonstrate here how jointly learning both of them can be fruitful. In this paper, we propose RumorSleuth, a multitask deep learning model which can leverage both the textual information and user profile information to jointly identify the veracity of a rumor along with users' stances. Tests on two publicly available rumor datasets demonstrate that RumorSleuth outperforms current state-of-the-art models and achieves up to 14% performance gain in rumor veracity classification and around 6% improvement in user stance classification.

Cyber-physical systems (CPS) are ubiquitous in several critical infrastructure applications. Forecasting the state of CPS, is essential for better planning, resource allocation and minimizing operational costs. It is imperative to forecast the state of a CPS multiple steps into the future to afford enough time for planning of CPS operation to minimize costs and component wear. Forecasting system state also serves as a precursor to detecting process anomalies and faults. Concomitantly, sensors used for data collection are commodity hardware and experience frequent failures resulting in periods with sparse or no data. In such cases, re-construction through imputation of the missing data sequences is imperative to alleviate data sparsity and enable better performance of down-stream analytic models. In this paper, we tackle the problem of CPS state forecasting and data imputation and characterize the performance of a wide array of deep learning architectures - unidirectional gated and non-gated recurrent architectures, sequence to sequence (Seq2Seq) architectures as well as bidirectional architectures - with a specific focus towards applications in CPS. We also study the impact of procedures like scheduled sampling and attention, on model training. Our results indicate that Seq2Seq models are superior to traditional step ahead forecasting models and yield an improvement of at least 28.5% for gated recurrent architectures and about 87.6% for non-gated architectures in terms of forecasting performance. We also notice that bidirectional models learn good representations for forecasting as well as for data imputation. Bidirectional Seq2Seq models show an average improvement of 17.6% in forecasting performance over their unidirectional counterparts. We also demonstrate the effect of employing an attention mechanism in the context of Seq2Seq architectures and find that it provides an average improvement of 57.12% in the case of unidirectional Seq2Seq architectures while causing a performance decline in the case of bidirectional Seq2Seq architectures. Finally, we also find that scheduled sampling helps in training better models that yield significantly lower forecasting error.

Nowadays social network platforms like Twitter, Facebook, Weibo have created a new landscape to communicate with our friends and the world at large. In this landscape our social activities, purchase decisions, check-ins etc. become available immediately to our friends/followers and thus encouraging them to involve in the same activity. This gives rise to the question, given a user and her friends' previous actions, can we predict what is she going to do next? This problem can serve as a good indicator enabling policy research, targeted advertising, assortment planning etc. To capture such sequential mechanism two broad classes of methods have been proposed in the past. First one is the Markov Chain (MC), which assumes user's next action can be predicted based on her most recently taken actions while the second type of approach i.e. Recurrent Neural Network (RNN) tries to model both long and short term preferences of a user. However, none of the two classes of models contain any integrated mechanism to capture the preferences of neighbor's actions. To fill this gap, we propose a social network augmented neural network model named NActSeer which takes the neighbors' actions into account in addition to the user's history. To achieve this NActSeer maintains a dynamic user embedding based on the activities within a time window. It then learns a feature representation for each user which is augmented by her neighbors. Empirical studies on four real-world datasets show that NActSeer is able to outperform several classical and state-of-the-art models proposed for similar problems and achieves up to 71% performance boost.

Multivariate time series forecasting is an important task in state forecasting for cyber-physical systems (CPS). State forecasting in CPS is imperative for optimal planning of system energy utility and understanding normal operational characteristics of the system thus enabling anomaly detection. Forecasting models can also be used to identify sub-optimal or worn out components and are thereby useful for overall system monitoring. Most existing work only performs single step forecasting but in CPS it is imperative to forecast the next sequence of system states (i.e curve forecasting). In this paper, we propose DyAt (Dynamic Attention) networks, a novel deep learning sequence to sequence (Seq2Seq) model with a novel hierarchical attention mechanism for long-term time series state forecasting. We evaluate our method on several CPS state forecasting and electric load forecasting tasks and find that our proposed DyAt models yield a performance improvement of at least 13.69% for the CPS state forecasting task and a performance improvement of at least 18.83% for the electric load forecasting task over other state-of-the-art forecasting baselines. We perform rigorous experimentation with several variants of the DyAt model and demonstrate that the DyAt models indeed learn better representations over the entire course of the long term forecast as compared to their counterparts with or without traditional attention mechanisms. All data and source code has been made available online.

Mass gatherings often underlie civil disobedience activities and as such run the risk of turning violent, causing damage to both property and people. While civil unrest is a rather common phenomenon, only a small subset of them involve crowds turning violent. How can we distinguish which events are likely to lead to violence? Using articles gathered from thousands of online news sources, we study a two-level multi-instance learning formulation, CrowdForecaster, tailored to forecast violent crowd behavior, specifically violent protests. Using data from five countries in Latin America, we demonstrate not just the predictive utility of our approach, but also its effectiveness in discovering triggering factors, especially in uncovering how and when crowd behavior begets violence.

Cascades are an accepted model to capturing how information diffuses across social network platforms. A large body of research has been focused on dissecting the anatomy of such cascades and forecasting their progression. One recurring theme involves predicting the next stage(s) of cascades utilizing pertinent information such as the underlying social network, structural properties of nodes (e.g., degree) and (partial) histories of cascade propagation. However, such type of granular information is rarely available in practice. We study in this paper the problem of cascade prediction utilizing only two types of (coarse) information, viz. which node is infected and its corresponding infection time. We first construct several simple baselines to solve this cascade prediction problem. Then we describe the shortcomings of these methods and propose a new solution leveraging recent progress in embeddings and attention models from representation learning. We also perform an exhaustive analysis of our methods on several real world datasets. Our proposed model outperforms the baselines and several other state-of-the-art methods.

EMBERS is an anticipatory intelligence system forecasting population-level events in multiple countries of Latin America. A deployed system from 2012, EMBERS has been generating alerts 24x7 by ingesting a broad range of data sources including news, blogs, tweets, machine coded events,currency rates, and food prices. In this paper, we describe our experiences operating EMBERS continuously for nearly 4 years, with specific attention to the discoveries it has enabled, correct as well as missed forecasts, lessons learnt from participating in a forecasting tournament, and our perspectives on the limits of forecasting including ethical considerations.

Forecasting large-scale societal events like civil unrest movements, disease outbreaks, and elections is an important and challenging problem. From the perspective of human analysts and policy makers, forecasting algorithms must not only make accurate predictions but must also provide supporting evidence, e.g., the causal factors related to the event of interest. We develop a novel multiple instance learning based approach that jointly tackles the problem of identifying evidence-based precursors and forecasts events into the future. Specifically, given a collection of streaming news articles from multiple sources we develop a nested multiple instance learning approach to forecast significant societal events such as protests. Using data from three countries in Latin America, we demonstrate how our approach is able to consistently identify news articles considered as precursors for protests. Our empirical evaluation demonstrates the strengths of our proposed approach in filtering candidate precursors, in forecasting the occurrence of events with a lead time advantage and in accurately predicting the characteristics of civil unrest events.

Civil unrest (protests, strikes, and “occupy” events) is a common occurrence in both democracies and authoritarian regimes. The study of civil unrest is a key topic for political scientists as it helps capture an important mechanism by which citizenry express themselves. In countries where civil unrest is lawful, qualitative analysis has revealed that more than 75% of the protests are planned, organized, and/or announced in advance; therefore detecting future time mentions in relevant news and social media is a direct way to develop a protest forecasting system. We develop such a system in this paper, using a combination of key phrase learning to identify what to look for, probabilistic soft logic to reason about location occurrences in extracted results, and time normalization to resolve future tense mentions. We illustrate the application of our system to 10 countries in Latin America, viz. Argentina, Brazil, Chile, Colombia, Ecuador, El Salvador, Mexico, Paraguay, Uruguay, and Venezuela. Results demonstrate our successes in capturing significant societal unrest in these countries with an average lead time of 4.08 days. We also study the selective superiorities of news media versus social media (Twitter, Facebook) to identify relevant tradeoffs.

In recent years, the amount of information shared (both implicit and explicit) between traditional news media and social media sources like Twitter has grown at a prolific rate. Traditional news media is dependent on social media to help identify emerging developments; social media is dependent on news media to supply information in certain categories. In this paper, we present a principled framework for understanding their symbiotic relationship, with the goal of (1) understanding the type of information flow between news articles and the Twitterverse by classifying it into four states; (2) chaining similar news articles together to form story chains and extracting interaction patterns for each story chain in terms of interaction states of news articles in the story chain, and (3) identifying major interaction patterns by clustering story chains and understanding their differences by identifying main topics of interest within such clusters.

Developed under the IARPA Open Source Initiative program, EMBERS (Early Model Based Event Recognition using Surrogates) is a large-scale Big-Data analytics system for forecasting significant societal events, such as civil unrest incidents and disease outbreaks on the basis of continuous, automated analysis of large volumes of publicly available data. It has been operational since November of 2012, delivering approximately 50 predictions each day. EMBERS is built on a streaming, scalable, share-nothing architecture and is deployed on Amazon Web Services (AWS).

We describe the design, implementation, and evaluation of EMBERS, an automated, 24x7 continuous system for forecasting civil unrest across 10 countries of Latin America using open source indicators such as tweets, news sources, blogs, economic indicators, and other data sources. Unlike retrospective studies, EMBERS has been making forecasts into the future since Nov 2012 which have been (and continue to be) evaluated by an independent T&E team (MITRE). Of note, EMBERS has successfully forecast the June 2013 protests in Brazil and Feb 2014 violent protests in Venezuela. We outline the system architecture of EMBERS, individual models that leverage specific data sources, and a fusion and suppression engine that supports trading off specific evaluation criteria. EMBERS also provides an audit trail interface that enables the investigation of why specific predictions were made along with the data utilized for forecasting. Through numerous evaluations, we demonstrate the superiority of EMBERS over baserate methods and its capability to forecast significant societal happenings.