Research Pillar: Human-Centered AI

Artificial intelligence is often perceived as a competition to human intelligence, however, a more likely scenario for the future digital society is one where there is collaboration between humans and AI. This collaboration is an active research area in computer science and social science, with work on interactive learning, trust in AI, and explainability of machine decisions. One key area of study is how people interact with AI, specifically speech-based interactive technologies like virtual assistants. Researchers are analyzing user behavior and perceptions of these systems, observing individuals as they use AI in their daily lives, and using data science methods to examine online resources such as reviews and opinion pieces. The research questions that are being addressed include "Can we develop an automated approach to understand user sentiment and perceptions towards AI-based systems, specifically in the context of smart speakers?", "What are the most common misconceptions about AI and how do they impact user interactions with the technology?" and "Can we develop an automatic method for identifying and addressing user's misconceptions about AI-based systems?". Additionally, the integration of AI into information systems and collaboration platforms can enhance human abilities and create collective intelligence in mixed teams of humans and AI-driven agents. It is important to consider the ethical and social implications of human-AI cooperation, as well as the potential impact on employment and the economy.

Projects:

• Industry Project with Honda Research Institute (HRI)
• MOTIV
• AIL at Work

Principal Investigators:

• Marc Latoschik
• Andreas Hotho
• Carolin Wienrich
• Ingo Scholtes

Publications:

• Willing to Revise? Confidence and Recommendation Adoption in AI-Assisted Image Recognition.
  In: HHAI2022: Augmenting Human Intellect. IOS Press, 2022.
  Leonore Röseler, Ingo Scholtes, Bernhard Sendhoff and Anikó Hannák.
  [doi]  [BibTeX] 

The increasing volume of available data on social systems opens new opportunities for large-scale, quantitative studies of social phenomena. Such studies can help us to better understand how humans communicate and collaborate, what makes teams productive, what mechanism are at work in successful social organizations, and how technology shapes human behavior. This research not only offers new ways to address long-standing issues in the social sciences, it is also crucial to model, design and manage socio-technical systems.

Addressing these questions, several groups at CAIDAS use data science and machine learning techniques to study social organizations. In a large-scale analysis of data on more than 30,000 developers in 58 Open Source Software projects, we could validate and quantify the Ringelmann effect known from social psychology and organizational theory. We could also show how coordination structures in software development teams influence the productivity of team members. Studying large bibliographic data sets, we could further show how social mechanisms influence editorial processes and citation practices. These works provide actionable insights for project management and policy-making.

Principal Investigators:

• Ingo Scholtes
• Goran Glavaš

Publications:

• Predicting scientific success based on coauthorship networks..
  EPJ Data Sci., 3(1):9, 2014.
  Emre Sarigöl, René Pfitzner, Ingo Scholtes, Antonios Garas and Frank Schweitzer.
  [doi]  [BibTeX] 

Software systems are at the heart of the digital society: They control critical infrastructures like communication or energy systems, fuel the increasing automation in industrial manufacturing and are key drivers of the digital economy. Despite this importance, the development of complex software systems is still a fundamental challenge. Credible reports indicate that the majority of software projects run over time or budget -- or fail altogether, resulting in billions of dollars wasted every year. And while technical aspects like, e.g., programming techniques, testing methods, or developer support tools have improved significantly over the past years, our understanding how human and social factors contribute to success or failure of software projects is still in its infancy.

Addressing these challenges, we use data science to quantitatively study collaborative software engineering processes. As an example, we use network analysis and statistical modeling to study the evolution of software architectures based on large-scale data from software repositories. This not only allows us to trace the maintainability of software systems. We can also assist developers in the refactoring of code. We further extract large data sets from online support tools, and analyze them to better understand how social factors influence software development processes. This approach has helped us to uncover social mechanisms at work in software development, to quantify risks in Open Source communities, and to improve information systems used by software development teams.

Principal Investigator:

• Ingo Scholtes

Publications:

• Categorizing bugs with social networks: A case study on four open source software communities.
  In: 2013 35th International Conference on Software Engineering (ICSE), pages 1032-1041. 2013.
  Marcelo Serrano Zanetti, Ingo Scholtes, Claudio Juan Tessone and Frank Schweitzer.
  [doi]  [abstract]  [BibTeX] 

• The rise and fall of a central contributor: Dynamics of social organization and performance in the GENTOO community.
  In: 2013 6th International Workshop on Cooperative and Human Aspects of Software Engineering (CHASE), pages 49-56. 2013
  Marcelo Serrano Zanetti, Ingo Scholtes, Claudio Juan Tessone and Frank Schweitzer.
  [doi]  [abstract]  [BibTeX] 

This research area focuses on making state-of-the-art language technology accessible to a wider range of people, in particular speakers of low-resource languages and those with limited access to large-scale computational resources. CAIDAS primarily uses deep learning and representation learning methods for semantic modeling of natural language, with a focus on multilingual representation learning and cross-language transfer of models for specific NLP tasks. In addition to this core line of work on multilinguality, CAIDAS researchers also address challenges of sustainability, that is, reducing the carbon footprint of NLP research through more efficient training of language models and fairness, that is, removal of negative societal stereotypes and biases from language models. Finally, we enable researchers in social sciences and humanities as well as practitioners from various application areas to benefit from cutting-edge NLP models through interdisciplinary projects. In sum, CAIDAS develops sustainable, modular and computationally efficient NLP models, fair and ethical NLP, and truly multilingual NLP with a focus on low-resource languages.

Principial Investigator:

• Goran Glavaš

Publications:

• Political Text Scaling Meets Computational Semantics.
  ACM/IMS Transactions on Data Science, 2(4):1-27, 2021.
  Federico Nanni, Goran Glavaš, Ines Rehbein, Simone Paolo Ponzetto and Heiner Stuckenschmidt.
  [doi]  [BibTeX] 

Recommender systems support users in their decisions by suggesting personalized and suitable recommendations in various domains. Latest research focus is on deep learning models for the sequential user behaviour, e.g. for customer in online shops or in online games. Content based recommender focus on the integration of formalized knowledge from ontologies and unstructured information such as natural language, to improve the treatment of patients in hospitals.

Projects:

• industry cooperation project with adidas
• DZ.PTM

Principal Investigator:

• Andreas Hotho

Publications:

• A Case Study on Sampling Strategies for Evaluating Neural Sequential Item Recommendation Models.
  In: Fifteenth ACM Conference on Recommender Systems. ACM, 2021.
  Alexander Dallmann, Daniel Zoller and Andreas Hotho.
  [BibTeX]