Papers by Ronald Cardenas

5 papers
Document Modeling with External Attention for Sentence Extraction (P18-1)

Copied to clipboard

Challenge: Document modeling is essential to a variety of natural language understanding tasks.
Approach: They propose to use external information to improve document modeling for sentence extraction problems.
Outcome: The proposed model outperforms baseline models on document summarization and answer selection tasks and achieves state-of-the-art results on WikiQA and NewsQA.
A Grounded Unsupervised Universal Part-of-Speech Tagger for Low-Resource Languages (N19-1)

Copied to clipboard

Challenge: Unsupervised part of speech (POS) tagging is often framed as a clustering problem, but taggers need to ground their clusters as well.
Approach: They propose an approach for low-resource unsupervised part of speech (POS) tagging that yields fully grounded output and requires no labeled training data.
Outcome: The proposed method achieves reasonable performance across languages, including Sinhalese and Kinyarwanda, with no labeled training data.
GEMv2: Multilingual NLG Benchmarking in a Single Line of Code (2022.emnlp-demos)

Copied to clipboard

Challenge: Evaluations in machine learning rarely use the latest metrics, datasets, or human evaluation in favor of remaining compatible with prior work.
Approach: They propose to use the Generation, Evaluation, and Metrics Benchmark to integrate new evaluation methods into existing evaluations.
Outcome: The proposed evaluation infrastructure bridges the gap between the advantages of leaderboards and in-depth and evolving evaluations by allowing model developers to benefit from each other's work.
‘Don’t Get Too Technical with Me’: A Discourse Structure-Based Framework for Automatic Science Journalism (2023.emnlp-main)

Copied to clipboard

Challenge: Science journalism is the production of journalistic content covering scientific topics that are not covered in the scientific literature.
Approach: They propose to use a dataset to generate a scientific paper's tuples, a summary snippet and a novel technical framework to integrate a paper' s discourse structure with its metadata to guide generation.
Outcome: The proposed system outperforms baseline methods in elaborating a content plan meaningful for the target audience, simplifying the information selected, and producing a coherent final report in a layman’s style.
SparsePO: Controlling Preference Alignment of LLMs via Sparse Token Masks (2025.findings-emnlp)

Copied to clipboard

Challenge: Current direct preference optimization algorithms focus on a strict set of tokens contributing signals of KL divergence and rewards to the loss function.
Approach: They propose a method that automatically learns to weight the KL divergence and reward corresponding to each token during PO training.
Outcome: The proposed method achieves +10% and +3% win-rate points in two PO scenarios.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations