Riset Pengalaman Pengguna untuk Perancangan Fitur Peringkas Literatur Berbasis AI pada Repositori Institusi

Abstract

Pertumbuhan volume publikasi ilmiah yang pesat di repositori institusi memicu information overload yang menghambat efisiensi mahasiswa dalam tinjauan pustaka. Di IPB Scientific Repository, keterbatasan metadata dan abstrak yang tidak konsisten mendorong involuntary PDF download dan screening fatigue. Penelitian ini bertujuan melakukan riset pengalaman pengguna sebagai dasar perancangan fitur peringkas literatur berbasis AI pada repositori institusi. Metode penelitian menggunakan pendekatan Design Thinking pada tiga fase: Empathize, Define, dan Ideate. Data dikumpulkan melalui kuesioner screening terhadap 41 responden, serta wawancara mendalam dan card sorting bersama enam partisipan mahasiswa IPB. Analisis menggunakan thematic analysis dan analisis deskriptif. Hasil mengonfirmasi information overload: 90,2% responden sering mengunduh PDF lengkap akibat abstrak tidak memadai, dan 78,0% sering kehilangan jejak dokumen. Thematic analysis menghasilkan 24 tema UX Insight yang dikristalisasi menjadi lima dimensi Point of View dan diterjemahkan menjadi sembilan pertanyaan How Might We sebagai peta peluang solusi. Prioritization Matrix mengidentifikasi 17 ide solusi prioritas (High Impact/Low Effort), didominasi fitur ekstraksi konten otomatis, sitasi interaktif, dan format toggle ringkasan. Artefak pradesain ini menjadi cetak biru konseptual bagi pengembangan sistem repositori cerdas yang diarahkan untuk mengurangi beban kognitif pengguna.

The rapid growth of scientific publications in institutional repositories has triggered information overload, hindering students' efficiency in literature reviews. At the IPB Scientific Repository, inconsistent metadata and abstract quality force students into involuntary PDF downloads and screening fatigue. This study conducts user experience research to inform the design of an AI-based literature summarization feature for institutional repositories. The research employs a Design Thinking approach across three phases: Empathize, Define, and Ideate. Data were collected through a screening questionnaire (41 respondents) and in-depth interviews with card sorting (six IPB students). Data analysis utilized thematic analysis and descriptive analysis. The findings confirm the information overload phenomenon: 90.2% of respondents frequently download full PDFs due to inadequate abstracts, and 78.0% frequently lose track of opened documents. Thematic analysis produced 24 UX Insight themes, crystallized into five Point of View dimensions and translated into nine How Might We questions as a solution opportunity map. The Prioritization Matrix identifies 17 high-priority solution ideas (High Impact/Low Effort), dominated by automatic content extraction, interactive citations, and summary format toggle features. These pre-design artifacts serve as a blueprint for developing a smart repository system aimed at reducing users cognitive load.