Papers by Steven Skiena
Evaluating Language Translation Models by Playing Telephone (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing language models are inadequate for evaluating machine translation systems . current evaluation methods are costly and require specialized expertise to prepare and score gold standard translations . |
| Approach: | They propose an unsupervised method to generate training data for translation evaluation by repeated rounds of translation between source and target languages. |
| Outcome: | The proposed method outperforms a popular translation evaluation system on two tasks . human annotation is costly and requires specialized expertise to prepare and score gold standard translations . |
GNAT: A General Narrative Alignment Tool (2023.emnlp-main)
Copied to clipboard
| Challenge: | Algorithmic sequence alignment is a common operation in many NLP tasks, but it is difficult to recognize similarities between distant versions of narratives such as translations and retellings. |
| Approach: | They propose a general approach to narrative alignment coupling the Smith-Waterman algorithm from bioinformatics with modern text similarity metrics. |
| Outcome: | The proposed approach can be used to identify similarities between two different versions of narratives, and to define rigorous p-values on the significance of any alignment. |
The Shape of Word Embeddings: Quantifying Non-Isometry with Topological Data Analysis (2024.findings-emnlp)
Copied to clipboard
| Challenge: | a recent study shows that word embeddings represent language vocabularies as clouds of d-dimensional points . authors assume that word embedded in different languages are essentially isometric . |
| Approach: | They use persistent homology to measure distances between language pairs from unlabeled embeddings . they construct language phylogenetic trees over 81 Indo-European languages . |
| Outcome: | The proposed tree shows that the embeddings differ from the reference tree. |
Analyzing Film Adaptation through Narrative Alignment (2023.emnlp-main)
Copied to clipboard
| Challenge: | a new study examines the book-to-film adaptation process by examining the differences between the two media . novel adaptations often require dropping sections of the source text from the movie script . |
| Approach: | They use a Smith-Waterman local alignment algorithm to quantify text similarity between scenes and book units. |
| Outcome: | The proposed method reveals that novel adaptations often require dropping parts of the source text from the movie script. |
Learning and Evaluating Character Representations in Novels (2022.findings-acl)
Copied to clipboard
| Challenge: | Recent advances in word embeddings have proven successful in learning entity representations from short texts but do not capture full book-level information. |
| Approach: | They propose two novel ways to learn fixed-length vector representations of characters from novels . they use graph neural network-based embeddings from a full corpus-based character network . |
| Outcome: | The proposed methods outperform text-based embeddings in four tasks. |