Graduation Semester and Year
Fall 2025
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Computer Science
Department
Computer Science and Engineering
First Advisor
Chengkai Li
Abstract
False information—whether shared unintentionally as misinformation or deliberately crafted as disinformation—poses significant risks in critical domains such as public health, politics, and science. Among the most pervasive disinformation techniques is cherry-picking, in which selectively presented evidence creates distorted or misleading narratives that can undermine trust and hinder sound decision-making. At the same time, the broader fact-checking process remains challenged by the need to efficiently gather and evaluate vast amounts of evidence. This thesis addresses both challenges through three complementary contributions. First, it addresses selective evidence presentation by introducing a novel approach to automatically detecting cherry-picked statements by identifying missing important statements in a target news story using contextual information from other news sources with different biases. This thesis showcases the flexibility of the proposed approach by utilizing different models, including fine-tuned embedding models, zero and few-shot generative models, in addition to unsupervised models. This research introduces a novel dataset specifically designed for training and evaluating cherry-picking detection models. The best performing model achieves an F-1 score of about 89% in detecting important statements. Moreover, results show the effectiveness of incorporating external knowledge from alternative narratives when assessing statement importance. Second, this thesis targets the problem of curating high quality evidence for information verification by presenting ClaimMap, an automated evidence curation and modeling tool designed to support the fact-checking process. ClaimMap helps fact-checkers prioritize high-quality, useful evidence and visualizes the claim, evidence, and their relationships through an interactive map. ClaimMap involves three aspects in prioritizing evidence for fact-checking, including content, context, and source. Our evaluations show that ClaimMap achieves solid performance on our dataset of 120 PolitiFact claims and performs comparably to existing methods on the external AVeriTeC benchmark. Third, this thesis presents an open-source dashboard designed to monitor and mitigate the COVID-19 misinfodemic by integrating epidemiological data, large-scale social media analysis, and a curated catalog of verified facts. Using natural language processing techniques for claim matching and stance detection, the system enables real-time, spatiotemporal tracking of misinformation and public engagement with most prevalent facts across regions. The dashboard serves as an information surveillance and decision-support tool for researchers, public health officials, and the general public.
Keywords
cherry-picking, fact-checking, propaganda, manipulation, disinformation, misinformation
Disciplines
Artificial Intelligence and Robotics | Data Science
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Jaradat, Israa, "FILLING THE BLANKS: CONTEXT-DRIVEN DETECTION OF CHERRY-PICKING IN NEWS REPORTING" (2025). Computer Science and Engineering Dissertations. 431.
https://mavmatrix.uta.edu/cse_dissertations/431