Christopher Klamm, Data and Web Science Group: Natural Language Processing (December 2022)

Christopher Klamm has been a Ph.D. student at the Chair of DWS of Prof. Dr. Simone Ponzetto since December 2021. His research focuses on analyzing social and political phenomena, such as populism, in texts using automatic methods of natural language processing. Previously, he studied computer science, political science, and philosophy in Darmstadt and Zurich.

What is your current research topic?       

In my research, I use automated text analysis to study the concept of populism and how it is represented in language. Populism is a widespread and multifaceted political phenomena that can vary significantly from party to party, context to context, and even over time. Populists often use rhetorical strategies and tactics to mobilize support among the general population, for example, by appealing to emotions, playing on people's fears, or framing issues in terms of a conflict between “the people” and “the elites”. Populism can be found on both the left and the right side of the political spectrum. This makes it challenging to define populism in a way that is both precise and captures its diversity. My work focuses on developing a measurable conceptualization of populism that can be used to identify it automatically in texts. This involves analyzing a large number of examples of populism that have been annotated by trained experts using an extended framework designed for this purpose. The goal is to teach machine learning models with these annotated examples to recognize the characteristics of populism in large datasets, using the German Bundestag parliamentary debates as a case study. By understanding how these models learn and whether they are able to capture complex social and political phenomena, we can gain insights into how populism emerges and how it can be detected and analyzed automatically.

For those who have not yet delved deeply into the topic of Data Science: How would you explain to a child what you are working on?

Populism is a special way of talking about politics that is based on the idea that the government should listen to the ordinary people and do what they want. People who use populist rhetoric often emphasize that they are fighting against a powerful group of elites who are trying to control the government. Populists might use loaded language, such as slogans, triggering words, catchphrases or moral aspects to try to rally people to their cause, and they often try to present themselves as champions and defenders of the common people. I study how populists use such language, like slogans and words about “the people” and “the elite” and teach computers to recognize these characteristics of populism so they can automatically identify it in political debates.

Everyone talks about Data Science – how would you describe the importance of the topic for yourself in three words?

Open science opportunity

What points of contact with Data Science does your work have? Which methods do you already use, and which would be interesting for you in the future?

I utilize a wide range of data science methods to analyze large datasets, such as parliamentary debates in Germany. My work involves training machine learning models in natural language processing to identify the individual dimensions and characteristics of populism in texts. These models aim to link specific patterns in language to properties of populism and to predict them automatically in texts like political speeches. For example, the models can learn to recognize when a politician speaks negatively about a group of elites in order to discredit them. The rapid development in automatic text analysis provides a rich foundation for future methodological improvements. In order to improve my analysis, I plan to utilize more domain-specific fine-grained language models that are able to capture the nuances and contextual features of language usage in different cultural settings and over time. By using data science techniques to capture a wider range of world knowledge, we can gain valuable insights into social and political phenomena and better understand their underlying mechanisms and dynamics.

How high is the value of Data Science for your work? Would your research even be possible without Data Science?

Data Science is absolutely crucial to my research, as it enables me to study populism and framing in language in a comprehensive way. Without the powerful tools of data science, I would not be able to analyze the large amounts of text data that are central to my work. In my research, data science is both a method and a touchstone to measure the effectiveness of new natural language processing methods and to analyze their abilities to capture complex concepts. Therefore, I also use data science to simultaneously analyze its limitations and highlight the need for further research to explore complex phenomena.

What development opportunities do you see for the topic of Data Science in relation to your field?

I see an incredible amount of development potential and can only briefly mention a few thoughts here. In the future, Data Science is likely to play an even more important role in computational social science research as researchers continue to take advantage of the vast amounts of data that are now available to them. This could include the use and combination of real-time data and the development of real-time analysis tools that enable researchers to more quickly and effectively analyze and respond to rapidly changing social phenomena. One key aspect of the development will be the adoption of open science practices, which involve making research data and methods more transparent and accessible to others. This will help to ensure that research findings are more reliable and can be more easily reproduced by others. Another important aspect of the future of Data Science in social science research will be the development of standards and best practices for processing data as well as training complex models on such data. However, bias will also be a key concern in the future of data science in social science research, as researchers seek to ensure that their data and methods are not biased in ways that could distort their findings. To address these issues, researchers may need to develop new benchmarks, specialized tests and other tools to help them assess the quality and reliability of their data and methods to gain more confidence in the new methods. However, I think that this development can only succeed if the various areas work closely together. I am looking forward to an exciting future.