An Exploration of Knowledge Editing for Arabic

TL;DR

This paper investigates knowledge editing techniques in Arabic, a morphologically rich language, evaluating various methods and extending LTE to multilingual settings, revealing challenges and improvements in cross-lingual generalization.

Contribution

First comprehensive study of Arabic knowledge editing, introducing multilingual LTE training and benchmarks, and analyzing method performances in cross-lingual contexts.

Findings

Parameter-based methods struggle with cross-lingual generalization.

Instruction-tuned methods show more robustness in Arabic KE.

Multilingual LTE training improves editability and transferability.

Abstract

While Knowledge Editing (KE) has been widely explored in English, its behavior in morphologically rich languages like Arabic remains underexamined. In this work, we present the first study of Arabic KE. We evaluate four methods (ROME, MEMIT, ICE, and LTE) on Arabic translations of the ZsRE and Counterfact benchmarks, analyzing both multilingual and cross-lingual settings. Our experiments on Llama-2-7B-chat show that parameter-based methods struggle with cross-lingual generalization, while instruction-tuned methods perform more robustly. We extend Learning-To-Edit (LTE) to a multilingual setting and show that joint Arabic-English training improves both editability and transfer. We release Arabic KE benchmarks and multilingual training for LTE data to support future research.