URIEL+: Enhancing Linguistic Inclusion and Usability in a Typological and Multilingual Knowledge Base

2024-09-27

International Conference on Computational Linguistics

AI Summary

URIEL+ addresses limitations in the URIEL typological knowledge base regarding language coverage breadth, completeness of typological features, and linguistic plausibility of distance computation. Methodologically, it systematically completes typological feature annotations for 2,898 previously underserved languages by integrating multi-source data from WALS, Glottolog, and Ethnologue; introduces a user-configurable distance metric framework that combines weighted cosine similarity with feature imputation, supporting three-dimensional vector representations—geographic, genealogical, and typological. Its contributions include constructing the most inclusive typological knowledge base to date (covering 4,005 languages), substantially improving alignment between computed cross-lingual distances and empirical linguistic conclusions; and achieving downstream task performance competitive with state-of-the-art baselines. URIEL+ thus provides robust, interpretable, and linguistically grounded cross-lingual distance estimates for multilingual modeling and language evolution research.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Knowledge Representation and Reasoning: Knowledge Representation Languages
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

URIEL is a knowledge base offering geographical, phylogenetic, and typological vector representations for 7970 languages. It includes distance measures between these vectors for 4005 languages, which are accessible via the lang2vec tool. Despite being frequently cited, URIEL is limited in terms of linguistic inclusion and overall usability. To tackle these challenges, we introduce URIEL+, an enhanced version of URIEL and lang2vec that addresses these limitations. In addition to expanding typological feature coverage for 2898 languages, URIEL+ improves the user experience with robust, customizable distance calculations to better suit the needs of users. These upgrades also offer competitive performance on downstream tasks and provide distances that better align with linguistic distance studies.

Problem

• Enhances linguistic inclusion in URIEL
• Improves usability of lang2vec tool
• Expands typological feature coverage

Innovation

• Expanded typological feature coverage
• Customizable distance calculations
• Improved linguistic distance alignment