Similarity between Units of Natural Language: The Transition from Coarse to Fine Estimation

TL;DR

This paper introduces a progressively refined regression model for measuring similarity between language units, improving accuracy and interpretability, especially in complex domains like legal and medical texts.

Contribution

It develops a novel similarity computation framework that iteratively refines measures and provides explanations, advancing beyond existing vague and hard-to-interpret methods.

Findings

Achieves state-of-the-art performance in edge case similarity measurement

Effectively identifies and addresses loopholes in similarity calculations

Enhances interpretability of similarity measures

Abstract

Capturing the similarities between human language units is crucial for explaining how humans associate different objects, and therefore its computation has received extensive attention, research, and applications. With the ever-increasing amount of information around us, calculating similarity becomes increasingly complex, especially in many cases, such as legal or medical affairs, measuring similarity requires extra care and precision, as small acts within a language unit can have significant real-world effects. My research goal in this thesis is to develop regression models that account for similarities between language units in a more refined way. Computation of similarity has come a long way, but approaches to debugging the measures are often based on continually fitting human judgment values. To this end, my goal is to develop an algorithm that precisely catches loopholes in a similarity calculation. Furthermore, most methods have vague definitions of the similarities they compute and are often difficult to interpret. The proposed framework addresses both shortcomings. It constantly improves the model through catching different loopholes. In addition, every refinement of the model provides a reasonable explanation. The regression model introduced in this thesis is called progressively refined similarity computation, which combines attack testing with adversarial training. The similarity regression model of this thesis achieves state-of-the-art performance in handling edge cases.