NLP for social data science
Natural Language Processing has many possible applications in social sciences. It can help to answer questions about sentiment, opinions, arguments, framing, toxicity, and many other subjects, enriching common social science research methods.
NLP methods are part and parcel of current research on social data science. SODAS researchers conducted a number of projects using a variety of NLP methods (particularly in analysis of social media). They also contribute to research on argument mining: an important task in fighting disinformation.
Social media are drawing global attention for the role they play in re-shaping society. Analysis of social media data is non-trivial: the data is noisy, highly contextual, ambiguous, and laden with privacy and copyright issues.
- Carlsen, H.A.B; Toubøl, Jonas; Ralund, S. (2021). ‘Brining Social Context Back In: Enrich Political Participation surveys with measures of social interaction from social media content data’. Public Opinion Quarterly. (Accepted)
- Carlsen, H. A. B., Toubøl, J., & Ralund, S. (2020). Consequences of Group Style for Differential Participation. Social Forces. https://doi.org/10.1093/sf/soaa063
- Ralund, S., Carlsen H.A.B, Lassen, David, D., Klemmensen, R. (2018) "What Topic Model Should I choose?" - measurement error induced by model choice in automated text analysis. Computational Sociology session at the American Sociological Association Annual Meeting.
Prior relevant work by the current SODAS staff:
- Rogers, A., Romanov, A., Rumshisky, A., Volkova, S., Gronas, M., & Gribov, A. (2018). RuSentiment: An Enriched Sentiment Analysis Dataset for Social Media in Russian. Proceedings of the 27th International Conference on Computational Linguistics, 755–763. http://aclweb.org/anthology/C18-1064
- Rogers, A., Kovaleva, O., & Rumshisky, A. (2019). Calls to Action on Social Media: Potential for Censorship and Social Impact. EMNLP-IJCNLP 2019 Second Workshop on Natural Language Processing for Internet Freedom. https://www.aclweb.org/anthology/D19-5005
Identifying claims and arguments in text is relevant for many social media analyses, and it is an important step towards fighting misinformation. If we can get an overview of the arguments people commonly make in relation to certain topics, we can then start to recognise the most common misperceptions as well as directly disinformative arguments.
Jakobsen, T. S. Th., Barrett, M., Søgaard, A. Spurious Correlations in Cross-Topic Argument Mining. Proceedings of *SEM 2021: The Tenth Joint Conference on Lexical and Computational Semantics. https://aclanthology.org/2021.starsem-1.25
A brief introductory video produced by SODAS on what Argument Mining is and why it is interesting: https://youtu.be/Xk-w_NZuU1k
|Search in Name||Search in Title||Search in Phone|
|Anna Rogers||Assistant Professor||+4535326548|
|Hjalmar Alexander Bang Carlsen||Assistant Professor||+4581936271|
|Terne Sasha Thorn Jakobsen||Research Assistant|
|Thyge Ryom Enggaard||Research Assistant|