The surge of reports from monitors like Amnesty International, Human Rights Watch, and the US State Department offers unprecedented detail on repression and rights protections—but most automatic approaches reduce texts to simple word counts that ignore syntax and word order, limiting what scholars and policy-makers can learn.
🔎 What Was Examined
- Human rights monitoring reports from major organizations (Amnesty International, Human Rights Watch, US State Department)
- The limits of conventional text representations that use lower-dimensional observations (e.g., word counts) and ignore syntax and word order
🛠️ How PULSAR Works
- A new system, PULSAR, processes text while taking syntax and word order into account
- Enables automated extraction of both the judgments expressed in a report and the specific aspects or rights those judgments concern
📈 Key Findings
- Detailed, syntax-aware representations improve predictions of two outcome domains tested: physical integrity rights and women’s political rights
- Models built with PULSAR are more interpretable than conventional specifications that rely on simple word counts
- The richer information extracted (judgments plus aspects/rights) enhances both predictive performance and clarity about what the model is using
🌐 Why It Matters
- Greater detail and interpretability create a coherent bridge between qualitative reading of human rights texts and large-scale quantitative analysis
- This approach expands what scholars and policy-makers can infer from the growing volume of human rights documentation and supports higher-resolution measurement of repression and protection
