Replicating Relevance-Ranked Synonym Discovery in a New Language and Domain

TL;DR

This paper adapts a relevance-ranked synonym discovery method to identify Swedish domain-specific synonyms in the building construction sector, demonstrating the effectiveness of learning to rank approaches and feature analysis in a new language and domain.

Contribution

It extends prior synonym discovery work to Swedish and construction domain, introducing new features and evaluating their impact with a learning to rank framework.

Findings

Learning to rank approach outperforms baseline methods

FastText embeddings serve as a strong baseline

Feature analysis reveals domain and language-specific differences

Abstract

Domain-specific synonyms occur in many specialized search tasks, such as when searching medical documents, legal documents, and software engineering artifacts. We replicate prior work on ranking domain-specific synonyms in the consumer health domain by applying the approach to a new language and domain: identifying Swedish language synonyms in the building construction domain. We chose this setting because identifying synonyms in this domain is helpful for downstream systems, where different users may query for documents (e.g., engineering requirements) using different terminology. We consider two new features inspired by the change in language and methodological advances since the prior work's publication. An evaluation using data from the building construction domain supports the finding from the prior work that synonym discovery is best approached as a learning to rank task in which a human editor views ranked synonym candidates in order to construct a domain-specific thesaurus. We additionally find that FastText embeddings alone provide a strong baseline, though they do not perform as well as the strongest learning to rank method. Finally, we analyze the performance of individual features and the differences in the domains.