ANALISIS SENTIMEN DAN PEMODELAN TOPIK PADA TWEET TERKAIT DATA BADAN PUSAT STATISTIK

Abstract

Annually, the Central Bureau of Statistics, known in Indonesia as BPS or Badan Pusat Statistik, conducts a routine Data Needs Survey (Survei Kebutuhan Data or SKD) to identify data requirements and the level of consumer satisfaction with the quality of data produced by BPS. However, SKD respondents are limited to consumers who have received services from the Integrated Statistics Services (Pelayanan Statistik Terpadu or PST) unit at BPS within a specific year. To gather opinions from the wider public accessing BPS data through channels other than the PST unit, an alternative approach is necessary – particularly through social media, specifically Twitter.
 This study employs Twitter data to analyze public sentiment regarding BPS data. To understand the distribution of topics discussed within the community about BPS data indicators, topic modeling has been employed. The sentiment analysis process utilizes IndoBERT, an Indonesian language Bidirectional Encoder Representations from Transformers (BERT) model. For topic modeling, the Latent Dirichlet Allocation (LDA) method is utilized.
 The results of sentiment analysis during the period 2020 - 2022 reveal that tweets related to BPS data generally convey a neutral sentiment. Meanwhile, the topic modeling process generates a range of topics, with variations observed in each year. Throughout 2020 - 2022, the most frequently discussed topics align with the statistical data from the 2020 - 2022 Data Needs Survey's data requirements section, reflecting the diversity of data needs.