Application of Fuzzy Search Method with Postgresql Trigram for Information Search Optimization on the Bojonegoro Regency PPID Website

Abdulloh Rozi
Aries Dwi Indriyanti

Jurnal Ilmiah Multidisiplin Indonesia
Vol. 5 No. 03 (2026): Jurnal Ilmiah Multidisplin Indonesia (JIM-ID), March 2026

The successful implementation of Law Number 14 of 2008 concerning Public Information Disclosure (Keterbukaan Informasi Publik - KIP) fundamentally relies on the public's ability to easily access information. The Official Information and Documentation Management Officer (Pejabat Pengelola Informasi dan Dokumentasi - PPID) Website of Bojonegoro Regency serves as the primary gateway for this service. However, the conventional search system (exact match) previously implemented proved to be rigid, often failing to retrieve documents due to minor typing errors (typos), thereby diminishing service efficiency. This research aims to design and implement a Fuzzy Search mechanism utilizing the PostgreSQL Trigram extension to replace the old system and to quantitatively evaluate its effectiveness. The Waterfall model was employed as the system development methodology, with implementation focusing on the backend using the PHP framework and a PostgreSQL database. A comparative experiment was conducted, contrasting the proposed system's performance against the existing exact match system , measured using standard Information Retrieval (IR) metrics: Precision, Recall, and F-Measure. The findings demonstrate that the existing exact match system showed a 100% failure rate when faced with common typing errors, whereas the Fuzzy Search implementation significantly increased error tolerance. Through empirical comparison, a Similarity Threshold configuration of 0.2 for PostgreSQL Trigram was identified as the optimal setting for the PPID Bojonegoro dataset. This value yielded the highest F-Measure score of 88.5%, reflecting the best balance between Recall (89%) and Precision (88%). Furthermore, the application of a GIN Index on PostgreSQL dramatically optimized performance, accelerating the search process by up to 40 times compared to the unindexed method. The proposed system successfully addresses the weakness against typos and ensures fast, inclusive public access to information, which is a key requirement under the KIP Law.