Yseop at FinSim-3 Shared Task 2021: Specializing Financial Domain Learning with Phrase Representations

TL;DR

This paper describes Yseop's approach to the FinSim-3 shared task, utilizing specialized sentence and word embeddings to classify financial terms into relevant hypernyms, achieving high accuracy and ranking second overall.

Contribution

The paper introduces a dual embedding approach combining Sentence-RoBERTa and FastText to improve financial term classification performance.

Findings

Achieved 0.917 average accuracy and 1.141 mean rank

Ranked 2nd overall in the shared task

Demonstrated effectiveness of specialized embeddings in financial NLP

Abstract

In this paper, we present our approaches for the FinSim-3 Shared Task 2021: Learning Semantic Similarities for the Financial Domain. The aim of this shared task is to correctly classify a list of given terms from the financial domain into the most relevant hypernym (or top-level) concept in an external ontology. For our system submission, we evaluate two methods: a Sentence-RoBERTa (SRoBERTa) embeddings model pre-trained on a custom corpus, and a dual word-sentence embeddings model that builds on the first method by improving the proposed baseline word embeddings construction using the FastText model to boost the classification performance. Our system ranks 2nd overall on both metrics, scoring 0.917 on Average Accuracy and 1.141 on Mean Rank.