Papers by Diana Maynard
TAMA: Target-Aware Multilingual Abuse Detection by Cascaded Conditional Multi-Task Learning (2026.acl-long)
Copied to clipboard
| Challenge: | Existing models for protecting public figures from online abuse ignore who is targeted and how. |
| Approach: | They propose a target-aware multi-task framework that conditions downstream predictions on upstream beliefs via three lightweight modules: Cross-Task Feature Fusion (CTF), Task-Adaptive Gating (TAG), and Label-Guided Span Detection (LGSD). |
| Outcome: | The proposed framework yields higher average F1 than single-task training and standard multi-task learning. |
Development of a Benchmark Corpus to Support Entity Recognition in Job Descriptions (2022.lrec-1)
Copied to clipboard
| Challenge: | Existing tools for identifying and extracting salient entities from job descriptions are limited by the lack of publicly available training data. |
| Approach: | They propose to use a standard definition of entities and a training corpus to develop a benchmark Entity Recognition (ER) model. |
| Outcome: | The proposed model achieves an F1 score of 0.59 from 18.6k entities comprising five types (Skill, Qualification, Experience, Occupation, and Domain). |
Dimensions of Online Conflict: Towards Modeling Agonism (2023.findings-emnlp)
Copied to clipboard
Matt Canute, Mali Jin, Hannah Holtzclaw, Alberto Lusoli, Philippa Adams, Mugdha Pandya, Maite Taboada, Diana Maynard, Wendy Hui Kyong Chun
| Challenge: | agonism fosters robust discussions, but hateful antagonism undermines constructive dialogue . a new study analyzes Twitter conversations to identify different dimensions of conflict . |
| Approach: | They annotated Twitter conversations related to trending controversial topics to model conflict on a richly annotized dataset. |
| Outcome: | The proposed model can help to moderate online conflicts and improve content monetization. |