Papers | Andy Halterman


Latent Civil War: Improving Inference and Forecasting with a Civil War Measurement Model

A new Bayesian measurement model for civil war and improved forecasting techniques.

Creating Custom Event Data Without Dictionaries: A Bag-of-Tricks

A bag of tricks for efficient, custom event data production, using transformer-based classifiers, question-answering models, Wikipedia entity linking, and active learning.

PLOVER and POLECAT: A New Political Event Ontology and Dataset

POLECAT is a new global event dataset for social science research, coded in the PLOVER event ontology.

Synthetically generated text for supervised text analysis

Synthetically generated text can help researchers address common issues in supervised text analysis.

Corpus-Level Evaluation for Event QA: The IndiaPoliceEvents Corpus Covering the 2002 Gujarat Violence

A corpus of complete annotations on police events in India with zero-shot QA baselines.

Few-Shot Upsampling for Protest Size Detection

New task and data on detecting the size of protests from news data with very limited training data.

The Case for Campaign Analysis: A Method for Studying Military Operations

Methodological guidance on conducting campaign analysis, a technique for studying military operations.

Extracting Political Events from Text Using Syntax and Semantics

A new technique for automatically extracting political events from text using grammatical structure and machine learning.

Violence against civilians in the Syrian civil war: Evidence from new micro-level data

New micro-level data on civilian casualities in Syria.

Geolocating Political Events in Text

A method for linking events with the place where they reportedly occur, using text.

Do the answers you get depend on the news you read? Protests and violence in Syria

Different text sources and classifiers produce different understanding of political violence in Syria

Word Order-Aware Text Processing: A Third Generation of Text-as-Data in Political Science

Dissertation prospectus framing three generations of text analysis in political science and prospects for word order-aware methods, especially for information extraction.

Mordecai: Full Text Geoparsing and Event Geolocation

Mordecai is an open source geoparser that can extract place names from text and resolve them to their geographic coordinates.

Securing Eastern Ukraine - Fighting Russia’s “Hybrid” Warfare

A military campaign analysis of Russian hybrid warfare in Ukraine and the logitical requirements for a hypothetical US military response.

Problems of Warning: Forecast Precision and Forecast Accuracy in Declassified State Department Cables

Learning warning signs from diplomatic cables and retrospective analysis of intelligence failures.