hariswb/analisa-konten

This repository is a small archive of Indonesian news analysis reports produced by an automated pipeline. The owner describes it as a place to keep static artifacts from AI agent work. The agent's job is to read articles from the Indonesian media and pull out four kinds of information: the sentiment of the coverage, the named entities mentioned, the relations between those entities, and the framing the article uses around a given topic. The analysis is organised around seven recurring Indonesian news topics. Legislasi covers proposed laws and judicial review. Demokrasi covers digital rights, press freedom, and criminalisation of speech. Lingkungan covers the energy transition, mining, and deforestation. Ekonomi covers the national budget, taxes, inflation, and poverty. Pasar Modal covers stocks, fintech, crypto, and banking. Hukum covers the courts, including the constitutional court and major verdicts. Keamanan covers the military, police, terrorism, and defence. The layout of the repo is straightforward. Inside a research folder, each report sits in its own dated subfolder named with the date and a topic slug. Each report folder has three things: a README.md with the analysis write-up, a charts folder containing the generated charts, and a data folder containing the source data used for the report. There is no application code in the visible README, only the convention for how these report folders are organised. The README is clear that the underlying article data does not originate in this repository. All of the analysis is powered by a separate Indonesian news monitoring platform called Semantik, at semantik.cc, which tracks and processes articles from national media outlets. The role of this repo, then, is to capture finished outputs, the charts and written analysis, in a form that can be browsed on GitHub without re-running anything. The project is new, with zero stars and a very short README. There are no installation instructions, no scripts, and no information about how the AI agent itself is set up or what tooling it uses. A reader interested in the data pipeline would need to look at the actual contents of the research folders for examples, or follow up with the Semantik platform that supplies the source articles.