Superlatives in Context: Modeling the Implicit Semantics of Superlatives

TL;DR

This work proposes a unified account of superlative semantics which allows for a broad-coverage annotation schema and annotated a multi-domain dataset of superlatives and their semantic interpretations, and specifically focuses on interpreting implicit or ambiguous superlative expressions, by analyzing how the discourse context restricts the set of interpretations.

Abstract

Superlatives are used to single out elements with a maximal/minimal property. Semantically, superlatives perform a set comparison: something (or some things) has the min/max property out of a set. As such, superlatives provide an ideal phenomenon for studying implicit phenomena and discourse restrictions. While this comparison set is often not explicitly defined, its (implicit) restrictions can be inferred from the discourse context the expression appears in. In this work we provide an extensive computational study on the semantics of superlatives. We propose a unified account of superlative semantics which allows us to derive a broad-coverage annotation schema. Using this unified schema we annotated a multi-domain dataset of superlatives and their semantic interpretations. We specifically focus on interpreting implicit or ambiguous superlative expressions, by analyzing how the discourse context restricts the set of interpretations. In a set of experiments we then analyze how well models perform at variations of predicting superlative semantics, with and without context. We show that the fine-grained semantics of superlatives in context can be challenging for contemporary models, including GPT-4.

Figures (6)

• Figure 1: An annotation example showing on the left the superlative (most), the sentence it appears in, and its previous context (shortened). On the right it shows the annotation slots (Target, DOI, CS etc.) and how they are filled given the text. Highlighted in yellow are the implicit discourse restrictions.
• Figure 2: Counts of the semantic types annotated over the different domains.
• Figure 3: Most frequent roles found in SuperSem.
• Figure 4: Most frequent properties in SuperSem.
• Figure 5: Accuracy for predicting the CS: given absolute vs. relative contexts. top_1: the first prediction in a beam is correct. top_5: at least one prediction in a beam of 5 is correct. comp/abs match: Does the type (absolute/relative) of the predicted CS fit the gold type?