Online Seminar Series of the Data Methods Initiative
The Data Methods Initiative (https://datamethodsinitiative.org/) is a collaborative academic group committed to promoting the use of computational methods (e.g., deep learning, natural language processing, computer vision) in political science, communications, internet research, and other disciplines. Next talk: "Towards Efficient and Accessible Geoparsing of Local Media: A Benchmark Dataset and LLM-based Approach" by Simona Bisiani (Surrey Institute for People-Centred AI) on October 1, 1:15 - 2:45 pm CET. This seminar introduces an innovative computational framework for extracting geographic references in local news, leveraging open-source, locally-run large language models. The first segment details the methodology, highlights its advantages over traditional approaches, and illustrates its application in identifying spatial patterns in news coverage-shedding light on media deserts and information inequalities across communities. A code demonstration will showcase the implementation using LMUK-Geo, a newly developed annotated corpus of UK local news articles designed for geoparsing evaluation. Attendees will leave with actionable knowledge to integrate these tools and methods into their research on local media landscapes and geographic trends. More info at https://datamethodsinitiative.org/
Earlier in the year, I edited a presentation of a somewhat similar effort at the US General services administration, entitled 'Public Pulse Mining'. As an alternative to formal surveys, they were using AI on scraped public data to calibrate reaction to construction projects. https://youtu.be/s6WDcbWIXck?list=PLEAvcNuSbEODSoH3wUFbMR8yl7SwwpsW4 joly On Tue, Sep 23, 2025 at 1:28 AM Marcel Garz via Air-L < air-l@listserv.aoir.org> wrote:
The Data Methods Initiative (https://datamethodsinitiative.org/) is a collaborative academic group committed to promoting the use of computational methods (e.g., deep learning, natural language processing, computer vision) in political science, communications, internet research, and other disciplines. Next talk: "Towards Efficient and Accessible Geoparsing of Local Media: A Benchmark Dataset and LLM-based Approach" by Simona Bisiani (Surrey Institute for People-Centred AI) on October 1, 1:15 - 2:45 pm CET. This seminar introduces an innovative computational framework for extracting geographic references in local news, leveraging open-source, locally-run large language models. The first segment details the methodology, highlights its advantages over traditional approaches, and illustrates its application in identifying spatial patterns in news coverage-shedding light on media deserts and information inequalities across communities. A code demonstration will showcase the implementation using LMUK-Geo, a newly developed annotated corpus of UK local news articles designed for geoparsing evaluation. Attendees will leave with actionable knowledge to integrate these tools and methods into their research on local media landscapes and geographic trends. More info at https://datamethodsinitiative.org/
_______________________________________________ The Air-L@listserv.aoir.org mailing list is provided by the Association of Internet Researchers http://aoir.org Subscribe, change options or unsubscribe at: http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
Join the Association of Internet Researchers: http://www.aoir.org/
-- -------------------------------------- Joly MacFie +12185659365 -------------------------------------- -
participants (2)
-
Joly MacFie -
Marcel Garz