A5041155583- Explainable Artificial Intelligence (XAI)
- Topic Modeling
- Forecasting Techniques and Applications
- Speech and dialogue systems
- Information Retrieval and Search Behavior
- AI in Service Interactions
- Data Quality and Management
- Advanced Text Analysis Techniques
- Radiomics and Machine Learning in Medical Imaging
- Semantic Web and Ontologies
- Adversarial Robustness in Machine Learning
- Research Data Management Practices
- Natural Language Processing Techniques
- Digital Marketing and Social Media
- Ethics and Social Impacts of AI
- Scientific Computing and Data Management
- Digital Communication and Language
- Web Data Mining and Analysis
- Health Sciences Research and Education
- Wikis in Education and Collaboration
- Biomedical Text Mining and Ontologies
- Teaching and Learning Programming
- Innovative Teaching and Learning Methods
- Electronic Health Records Systems
- Child and Animal Learning Development
University of Twente
2023-2024
University of Duisburg-Essen
2023
GESIS - Leibniz-Institute for the Social Sciences
2019-2022
Automated decision-making systems become increasingly powerful due to higher model complexity. While in prediction accuracy, Deep Learning models are black boxes by nature, preventing users from making informed judgments about the correctness and fairness of such an automated system. Explanations have been proposed as a general remedy box problem. However, it remains unclear if effects explanations on user trust generalise over varying accuracy levels. In online study with 959 participants,...
This report documents the program and outcomes of Dagstuhl Seminar 23031 "Frontiers Information Access Experimentation for Research Education", which brought together 38 participants from 12 countries. The seminar addressed technology-enhanced information access (information retrieval, recommender systems, natural language processing) specifically focused on developing more responsible experimental practices leading to valid results, both research as well scientific education. featured a...
Machine learning systems have become popular in fields such as marketing, financing, or data mining. While they are highly accurate, complex machine pose challenges for engineers and users. Their inherent complexity makes it impossible to easily judge their fairness the correctness of statistically learned relations between variables classes. Explainable AI aims solve this challenge by modelling explanations alongside with classifiers, potentially improving user trust acceptance. However,...
Abstract Purpose Publishing research data for reuse has become good practice in recent years. However, not much is known on how researchers actually find said data. In this exploratory study, we observe the information-seeking behaviour of social scientists searching to reveal impediments and identify opportunities search infrastructure. Methods We asked 12 participants observed them their natural environment. The sessions were recorded. Afterwards, conducted semi-structured interviews get a...
Conversational Agents are increasingly integrated into our daily routines, assisting us with various tasks, from simple commands such as scheduling events to more complex conversational search interactions. Such systems traditionally evaluated word-overlap metrics F1 score and accuracy. The full-day workshop on Search-Oriented Artificial Intelligence (SCAI) at CHIIR 2024 explored the evaluation of user's perspective. This interactive included multiple panel discussions working groups focused...
Supervised machine learning utilizes large datasets, often with ground truth labels annotated by humans. While some data points are easy to classify, others hard which reduces the inter-annotator agreement. This causes noise for classifier and might affect user's perception of classifier's performance. In our research, we investigated whether classification difficulty a point influences how strongly prediction mistake "perceived accuracy". an experimental online study, 225 participants...
Online retail has become a popular alternative to in-store shopping. However, unlike in traditional stores, users of online shops need find the right product on their own without support from expert salespersons. Conversational search could provide means compensate for shortcomings engines. To establish design guidelines such virtual assistants, we studied conversations user study (N = 24) where experts supported finding needs. We annotated concerning content and conversational structure...
With the emergence of voice assistants and large language models, conversational interaction with information has become part everyday life. The eighth edition search-oriented AI (SCAI) workshop brings together practitioners researchers from various disciplines to discuss challenges advances in search systems. This year's focuses on evaluations beyond relevance accuracy looks at user's perspective. features a shared task user-centered evaluation datasets metrics, challenging participants...
Abstract Publishing research data is widely expected to increase its reuse and inspire new research. In the social sciences, from surveys, interviews, polls, statistics are primary resources for There a long tradition collect offer in archives online repositories. Researchers use these systems identify relevant their However, especially search, users' complex information needs seem collide with capabilities of search systems. The capabilities, turn, depend high degree upon metadata schemes...
When talking about products, people often express their needs in vague terms with vocabulary that does not necessarily overlap product descriptions written by retailers. This poses a problem for chatbots online shops, as the vagueness and mismatch can lead to misunderstandings. In human-human communication, intuitively build common understanding throughout conversation, e.g., via feedback loops. To inform design of conversational search systems, we investigated effect different behaviors on...
With the rise of voice assistants and an increase in mobile search usage, natural language has become important query language. So far, most current systems are not able to process these queries because vagueness ambiguity Users have adapted their formulation what they think engine is capable of, which adds cognitive burden. our research, we contribute design interactive by investigating genuine information need a product scenario. In crowd-sourcing experiment, collected 132 needs We examine...
Shopping online is more and frequent in our everyday life. For e-commerce search systems, understanding natural language coming through voice assistants, chatbots or from conversational an essential ability to understand what the user really wants. However, evaluation datasets with detailed information needs of product-seekers which could be used for research do not exist. Due privacy issues competitive consequences, only few real queries logs are openly available. In this paper, we present...
Search systems on the Web rely user input to generate relevant results. Since early information retrieval systems, users are trained issue keyword searches and adapt language of system. Recent research has shown that often withhold detailed about their initial need, although they able express it in natural language. We therefore conduct a study (N = 139) investigate how four different design variants search interfaces can encourage reveal more information. Our results show chatbot-inspired...
Recent advances in natural language processing and deep learning have accelerated the development of digital assistants. In conversational commerce, these assistants help customers find suitable products online shops through conversations. During dialogue, assistant identifies customer's needs preferences subsequently suggests potentially relevant products. Traditional often allow users to filter search results based on their using facets. Selected facets can also serve as a reminder how...
Online retailers often offer a vast choice of products to their customers filter and browse through. The order in which the are listed depends on ranking algorithm employed online shop. State-of-the-art methods complex draw many different information, e.g., user query intent, product attributes, popularity, recency, reviews, or purchases. However, approaches that incorporate user-generated data such as click-through data, ratings, reviews disadvantage new have not yet been rated by...
Machine learning systems can help humans to make decisions by providing decision suggestions (i.e., a label for datapoint). However, individual datapoints do not always provide enough clear evidence confident suggestions. Although methods exist that enable identify those and subsequently abstain from suggesting label, it remains unclear how users would react such system behavior. This paper presents first findings user study on or labeling ambiguous datapoints. Our results show bear high...