Mind the Gap - Cross-Lingual Information Retrieval with Hierarchical Knowledge Enhancement - Abstract

Aug 26, 2026

Uploaded by Rowan

Cross-Lingual Information Retrieval (CLIR) ranks documents written in a different language from the user’s query, where language divergence creates a core difficulty. The work introduces a multilingual knowledge graph to leverage entity information across languages, enabling explicit query-document alignment and richer query representations. A model called HIKE encodes query, document, and KG text using multilingual BERT and fuses knowledge hierarchically to mitigate linguistic gaps, yielding significant improvements over prior methods.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Introduction

CLIR task and challenges

Knowledge-graph approach

Method: HIKE

Multilingual BERT encoding

Hierarchical information fusion (knowledge-level and language-level)

Experimental results

Mind the Gap: Cross-Lingual Information Retrieval with Hierarchical Knowledge Enhancement Fuwei Zhang 1, 2 , Zhao Zhang3, * , Xiang Ao 1, 2, 4 , Dehong Gao5 , Fuzhen Zhuang6, 7 , Yi Wei5 , Qing He 1, 2 1 Key Lab of Intelligent Information Processing of Chinese Academy of Sciences (CAS), Institute of Computing Technology, CAS, Beijing 100190, China 2 University of Chinese Academy of Sciences, Beijing 100049, China 3 Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China 4 Institute of Intelligent Computing Technology, Suzhou, CAS 5 Alibaba Group, Hangzhou, China 6 Institute of Artificial Intelligence, Beihang University, Beijing 100191, China 7 SKLSDE, School of Computer Science, Beihang University, Beijing 100191, China {zhangfuwei20g, zhangzhao2021, aoxiang, [heqing](heqing}@ict.ac.cn)[}](heqing}@ict.ac.cn)[@ict.ac.cn](heqing}@ict.ac.cn), [zhuangfuzhen@buaa.edu.cn](zhuangfuzhen@buaa.edu.cn), {dehong.gdh, [yi.weiy](yi.weiy}@alibaba-inc.com)[}](yi.weiy}@alibaba-inc.com)[@alibaba-inc.com](yi.weiy}@alibaba-inc.com) Abstract Cross-Lingual Information Retrieval (CLIR) aims to rank the documents written in a language different from the user’s query. The intrinsic gap between different languages is an essential challenge for CLIR. In this paper, we introduce the multilingual knowledge graph (KG) to the CLIR task due to the sufficient information of entities in multiple languages. It is regarded as a “silver bullet” to simultaneously perform explicit alignment between queries and documents and also broaden the representations of queries. And we propose a model named CLIR with hierarchical knowledge enhancement (HIKE) for our task. The proposed model encodes the textual information in queries, documents and the KG with multilingual BERT, and incorporates the KG information in the query-document matching process with a hierarchical information fusion mechanism. Particularly, HIKE first integrates the entities and their neighborhood in KG into query representations with a knowledge-level fusion, then combines the knowledge from both source and target languages to further mitigate the linguistic gap with a language-level fusion. Finally, experimental results demonstrate that HIKE achieves substantial improvements over state-ofthe-art competitors. Introduction The escalation of globalization burgeons the great demand for Cross-Lingual Information Retrieval (CLIR), which has broad applications such as cross-border e-commerce, crosslingual question answering, and so on (Li et al. 2020; R¨uckl, Swarnkar, and Gurevych 2019; Xu et al. 2021) . Informally, given a query in one language, CLIR is a document retrieval task that aims to rank the candidate documents in another language according to the relevance between the search query and the documents. * Corresponding author: Zhao Zhang Copyright © 2022, Association for the Advancement of Artificial Intelligence ([www.aaai.org](www.aaai.org)). All rights reserved. Most existing solutions to tackle the CLIR task are built upon machine translation (Dwivedi and Chandra 2016) systems (also known as MT systems) . One technical route is to translate either the query or the document to the same language as the other side (McCarley 1999; Aljlayl and Frieder 2001; Picchi and Peters 1998; Croft, Turtle, and Lewis 1991) . The other is to translate both the query and the document to the same intermediate language (Kishida and Kando 2003), e.g. English. After aligning the language of the query and documents, monolingual retrieval is performed to accomplish the task. Hence, the performance of the MT systems and the error accumulations may render them inefficient in CLIR. Recent studies strive to model CLIR with deep neural networks that encode both query and document into a shared space rather than using MT systems (Zhang et al. 2019; Sasaki et al. 2018; Hui et al. 2018a; Li et al. 2020) . Though these approaches achieve some remarkable success

FAQs

What does CLIR aim to do?

CLIR ranks candidate documents written in a language different from the query by measuring relevance between the query and documents.

Why is language gap a key challenge in CLIR?

Intrinsic differences between languages and limited explicit alignment in shared embedding approaches can reduce matching quality, especially when queries are short.

How does HIKE improve CLIR?

HIKE uses a multilingual knowledge graph and multilingual BERT, then injects KG information into query-document matching via hierarchical fusion: first at the knowledge level (entities and neighborhoods), then at the language level to further reduce the linguistic gap.

Aug 26, 2026

Uploaded by Rowan