Temporal Information Retrieval via Time-Specifier Model Merging

TL;DR

This paper introduces Time-Specifier Model Merging (TSM), a novel approach that improves temporal information retrieval by training specialized models for time constraints and merging them, enhancing temporal query performance without sacrificing non-temporal accuracy.

Contribution

The paper proposes TSM, a new method that merges specialized temporal retrievers to improve temporal IR while preserving non-temporal retrieval performance.

Findings

TSM significantly outperforms baseline methods on temporal queries.

TSM maintains high accuracy on non-temporal queries.

Extensive experiments validate the effectiveness of TSM.

Abstract

The rapid expansion of digital information and knowledge across structured and unstructured sources has heightened the importance of Information Retrieval (IR). While dense retrieval methods have substantially improved semantic matching for general queries, they consistently underperform on queries with explicit temporal constraints--often those containing numerical expressions and time specifiers such as ``in 2015.'' Existing approaches to Temporal Information Retrieval (TIR) improve temporal reasoning but often suffer from catastrophic forgetting, leading to reduced performance on non-temporal queries. To address this, we propose Time-Specifier Model Merging (TSM), a novel method that enhances temporal retrieval while preserving accuracy on non-temporal queries. TSM trains specialized retrievers for individual time specifiers and merges them in to a unified model, enabling precise handling of temporal constraints without compromising non-temporal retrieval. Extensive experiments on both temporal and non-temporal datasets demonstrate that TSM significantly improves performance on temporally constrained queries while maintaining strong results on non-temporal queries, consistently outperforming other baseline methods. Our code is available at https://github.com/seungyoonee/TSM .