Postdoctoral Researcher in Computational Social Science (NLP & Machine Learning)

The project

The ERC-funded project EMERGEPOL — “Emerging Issues and the Dynamics of Affective Polarization” — investigates how newly emerging political issues (such as immigration, climate change and European sovereignty) rise in salience, are framed by political elites and the media, and become drivers of affective polarization: the tendency of citizens to view political opponents with animosity and moral contempt. Adopting an “issue life-cycle” perspective, the project traces issues from elite and media discourse through to mass belief updating, ideological sorting and affective divides, using a most-different comparative design across four countries — Ireland, the Netherlands, Hungary and Türkiye.

Using a transnational and interdisciplinary approach, the project combines:

• Multilingual natural language processing (NLP) pipelines that detect and track the emergence, framing and moralization of issues across parliamentary debate, news media and social platforms;
• Three-wave panel surveys with embedded, pre-registered survey experiments, combined with Dynamic Bayesian Network (DBN) modelling to map the causal pathways from issue emergence to belief change and affective polarization;
• Machine-learning models that forecast the “polarization potential” of new issues, culminating in a Polarization Potential Index (PPI).

Your role

You will help build the multilingual text-as-data and forecasting backbone of EMERGEPOL (Project 1: detecting and tracking the emergence of new political issues; Project 3: forecasting their polarization potential). You will develop multilingual NLP pipelines that identify and track emerging political issues across elite discourse, news media and social platforms, and contribute to machine-learning models that forecast which issues are most likely to become polarizing. You will contribute to the project’s comparative and transnational framework alongside team members working on Ireland, the Netherlands, Hungary and Türkiye.

Research strands

• Multilingual NLP issue tracking (Project 1). Building and maintaining pipelines — transformer models (e.g. multilingual BERT, XLM-R), dynamic topic modelling, sentiment and moral-framing classification, and semantic-polarization scoring — over parliamentary debates, manifestos, news and social-media corpora in Turkish, Hungarian, Dutch and English.
• Network analysis and feature engineering. Constructing and analysing text-linked interaction/network data (graph-based NLP; coordination and amplification patterns) and engineering content, network and time-series features; managing large-scale data collection and access (platform APIs, archival collections, and requests under EU Digital Services Act Article 40) in line with ethics and GDPR requirements.
• Forecasting and the Polarization Potential Index (Project 3). Developing and validating machine-learning models that forecast issue-level “polarization potential” using leave-one-country-out validation, and contributing to the development of Polarization Potential Index (PPI) and its web-based tool for dissemination.

For this position, Koç University (Istanbul) will serve as your home base, with collaboration and, where relevant, research stays across the project’s country teams.

What do we expect from you?

• A PhD (awarded, or to be awarded before the start date) in computational social science, computer science, data science, computational linguistics, or political science with strong computational methods;
• Fluency in English (the project’s common working language);
• Demonstrated experience with NLP and/or machine learning applied to text (transformers, topic modelling, embeddings, or graph-based methods);
• Strong programming skills in Python and familiarity with relevant ML/NLP libraries;
• Experience working with large text corpora and building reproducible data pipelines;
• Sensitivity to research-ethics and data-protection (GDPR) requirements when working with media and social-media data.

You will:

• Conduct independent and collaborative research within an interdisciplinary, international team;
• Publish in peer-reviewed academic journals and contribute to collective project outputs;
• Present research at international conferences;
• Contribute to the organisation of workshops and public-facing events;
• Contribute to open-access pipelines, annotated corpora and codebases, and integrate NLP/ML outputs with the project’s panel-survey and DBN components.

Assets (not required, but considered an advantage):

• Working knowledge of one or two of the project’s research languages — Turkish, Hungarian, Dutch, or Irish/English — to support multilingual corpus development and validation;
• Experience with social-media data collection and platform APIs;
• Familiarity with graph neural networks or temporal/sequence models;
• Interest in political behaviour, public opinion, or polarization research.

What we offer

• A full-time position on a flagship ERC Advanced Grant. The initial appointment is for three years and, following a positive evaluation, is renewable for a further two years (up to five years in total);
• The planned start date is 1 January 2027, or as soon as possible thereafter;
• A competitive salary and social benefits in line with Koç University scales;
• A dedicated research budget for computing, data acquisition, fieldwork and international conference travel;
• A stimulating, international and interdisciplinary research environment spanning political behaviour, survey methodology, NLP, machine learning and Bayesian modelling, with close collaboration across the four country teams;
• Relocation and work-permit support for international candidates.

Want to apply?

You can apply for this vacancy through Koç University’s online application platform up to and including the application deadline (by midnight, Istanbul time). Please include the following attachments:

• Your CV, including a publication list and the contact details of two to three referees with references, websites and Github repositories;
• A motivation letter outlining your interest in the project and your relevant experience;
• A writing sample in English (e.g. a representative publication, dissertation chapter or academic paper);
• A short statement (max 2 pages, excluding references) indicating how you would approach your part of the EMERGEPOL project.

Informal enquiries about the position are welcome at acarkoglu@ku.edu.tr.

Selection process

• Step 1: initial selection based on the application file;
• Step 2: interview with shortlisted candidates (online or in person), which will include a short presentation of the candidate’s research statement aligned with the objectives of EMERGEPOL.

If you have questions about the job itself, please contact the Principal Investigator, Prof. Ali Çarkoğlu (acarkoglu@ku.edu.tr).

Koç University is committed to an inclusive working environment and equal opportunity. We welcome applications from candidates of all backgrounds and nationalities, regardless of gender identity, disability, ethnicity, religion or belief, sexual orientation or age.

Where to apply

E-mail: acarkoglu@ku.edu.tr

Requirements

• Research Field: Computer science » Informatics
• Education Level: PhD or equivalent
• Languages: ENGLISH
• Level: Excellent
• Research Field: Computer science » Informatics
• Years of Research Experience: 1 - 4

Work Location(s)

• Number of offers available: 1
• Company/Institute: Koç
• Country: Türkiye
• City: Istanbul

Contact

• City: Istanbul
• Website: https://www.ku.edu.tr/en/home
• Street: Sariyer
• Postal Code: 34450
• E-Mail: acarkoglu@ku.edu.tr