Neural Semantic Parsing with Extremely Rich Symbolic Meaning Representations

TL;DR

This paper introduces a hierarchical, taxonomical symbolic representation for semantic parsing that improves interpretability and out-of-vocabulary handling, compared to traditional methods, by leveraging lexical ontologies.

Contribution

It proposes a novel hierarchical symbolic representation and a neural taxonomical parser, enhancing semantic interpretability and OOV concept handling in open-domain semantic parsing.

Findings

Taxonomical model slightly underperforms standard model on traditional metrics.

Taxonomical model outperforms standard model on out-of-vocabulary concepts.

Rich hierarchical representations improve semantic parsing robustness.

Abstract

Current open-domain neural semantics parsers show impressive performance. However, closer inspection of the symbolic meaning representations they produce reveals significant weaknesses: sometimes they tend to merely copy character sequences from the source text to form symbolic concepts, defaulting to the most frequent word sense based in the training distribution. By leveraging the hierarchical structure of a lexical ontology, we introduce a novel compositional symbolic representation for concepts based on their position in the taxonomical hierarchy. This representation provides richer semantic information and enhances interpretability. We introduce a neural "taxonomical" semantic parser to utilize this new representation system of predicates, and compare it with a standard neural semantic parser trained on the traditional meaning representation format, employing a novel challenge set and evaluation metric for evaluation. Our experimental findings demonstrate that the taxonomical model, trained on much richer and complex meaning representations, is slightly subordinate in performance to the traditional model using the standard metrics for evaluation, but outperforms it when dealing with out-of-vocabulary concepts. This finding is encouraging for research in computational semantics that aims to combine data-driven distributional meanings with knowledge-based symbolic representations.