Sautex: A Language-Specific Phonetic Matching Algorithm for Resolving Spelling Variations in Hausa Personal Names

Spelling variations in personal names pose significant challenges for information retrieval and record linkage, particularly in low-resource languages such as Hausa. This paper presents a phonetic encoding algorithm, Sautex, specifically adapted to the phonological structure of Hausa, derived from the English Soundex system. Sautex was evaluated using a dataset of 17,591 Hausa name spelling attempts with edit distances ranging from 0 to 4. The system achieves a phonetic match accuracy of 77.20% and 66.86% recall for the positive class for the H* variant and 83.02% and 75.32% correspondingly for the H** variant, outperforming the baseline English Soundex by up to 11.53 and 16.76 percentage points in accuracy and recall for the positive class. These results demonstrate the viability of phonology-aware, language-specific encoding systems for African languages. Further studies might be undertaken to evaluate the performance of this algorithm on English names and its generalisation for other Nigerian names. The research aligns with two United Nations Sustainable Development Goals (SDGs), notably SDG 10 (Reduced Inequalities) by ensuring equitable digital representation of Hausa names, and SDG 9 (Industry, Innovation, and Infrastructure) by advancing localized NLP innovations.Note: i. Sautex is the contraction of the word Sauti, which means Sound in Hausa, and Soundex.ii. H* indicates values for the Sautex code with the first character includediii. H** indicates values for the Sautex code with the first character excludediv. English is abbreviated as Eng.