Export
Research Overview
Zeinab Abbassi focuses on algorithmic design for network-based systems, probabilistic modeling of information spread, and resource allocation in exchange markets. Her trajectory spans from offline matching approximations to optimal intervention strategies in epidemic models. Her most recent work addresses the optimization of display advertising within online social networks by leveraging social cues and network structure.
Research Trajectory
Abbassi’s research trajectory is characterized by the application of algorithmic and probabilistic frameworks to complex network environments, ranging from digital exchange markets to epidemiological models. Her early work focused heavily on the computational challenges of market design and resource distribution. As the first author of Offline matching approximation algorithms in exchange markets (2008), she investigated the problem of efficient offline matching within rapidly growing online social networks. This research formalized models for both one-shot markets and exchange markets over time, demonstrating that variants such as one-to-one exchanges and exchanges with short cycles are NP-hard. To address these computational bottlenecks, she proposed specific heuristics and approximation algorithms, showing that allowing exchanges through cycles significantly increases the volume of exchanged items. This work aligns with the broader area of The Computational Boundaries of Fair Resource Allocation, addressing how indivisible resources can be optimally distributed among agents.
Following her work on market matching, Abbassi shifted her focus toward network dynamics and the spread of contagions, applying probabilistic modeling to systemic vulnerabilities. In Toward optimal vaccination strategies for probabilistic models (2011), she led research examining how structural properties of networks can be leveraged to mitigate outbreaks. Recognizing that widespread disease transmission and the dissemination of computer viruses through email networks share a common underlying structure, this work addressed the logistical and economic constraints of universal vaccination. Abbassi and her co-author proposed strategies that utilize the properties of face-to-face social networks to identify critical individuals whose vaccination would most effectively reduce the overall infection rate. Around this period, her work also touched upon themes related to distributed systems and network bottlenecks, intersecting with the research direction Decoupling Data Dissemination from Ordering and Finality.
By 2015, Abbassi applied her expertise in probabilistic modeling and network structures to the domain of digital monetization. As the first author of Optimizing Display Advertising in Online Social Networks, she explored how conventional online advertising methods must be adapted to account for the distinct characteristics of social platforms. This research formalized the phenomenon of social cues—such as a user seeing that their friends have liked or clicked on an ad—and modeled the probability of a user's engagement as a function of their network's prior actions. By capturing these interdependent behaviors, the paper defined and addressed the resulting algorithmic challenges of optimizing display advertising in highly connected environments.
Current Directions
The most recent paper in the evidence is from 2015, and it indicates a focus on the intersection of algorithmic optimization and social influence within digital platforms. Recent work investigates how probabilistic models can accurately capture the effect of social cues on user engagement, particularly in the context of display advertising. By formalizing the relationship between an individual's likelihood of interacting with content and the prior actions of their network connections, this direction highlights the ongoing challenge of designing platform-specific algorithms that account for complex, interdependent user behaviors in online social networks.
Profile updated
Loading new papers…
Loading research neighbors…
See something wrong? Report a mistake