Search liu.se

Computational analysis offers new ways to derive meaning from text. We use large corpora of text as social sensors to measure what people feel, think, and talk about, which allows us to track the emergence of shared social understandings.

Researchers from sociology and statistics implement a scalable seeded topic model that extracts interpretable meaning structures in perhaps the largest text corpus ever analyzed in the social sciences.

We asked conservative and liberal Americans to evaluate true and false messages about politics, society, and science. The results reveal that people often genuinely believe fake news rather than pretending.

Analytical sociologists are harnessing troves of digitized text, digital trace, and social network data—along with the computational tools for their analysis—to answer sociology’s core questions in novel and rigorous ways.

As large corpora of digitized text become increasingly available, sociologists are rediscovering the potential of text data for inquiries into social and cultural phenomena.

An article shows how social scientists can leverage recent advances in natural language processing to automatically yet finely annotate millions of texts. They also provide a package and a tutorial for the open use of these methods.

In October 2019, we hosted scholars from across disciplines to discuss cutting-edge techniques for using large text corpora to understand how issues of migration and immigration are discussed in the political, media, and public spheres.

On October 2020, we hosted an online workshop to discuss issues and advances in performing text analysis on political, traditional media, and social media data.

We created 16 independent online ecosystems in which participants could share true and false messages about society, science, politics. Results reveal that partisan sorting systematically undermines the veracity of information circulating.

The growing size of many contemporary text corpora makes it increasingly difficult to use standard techniques of computational text analysis. Måns Magnusson and colleagues address these challenges in two papers.