项目介绍

2024年10月25日

The Interactive Visual Data Analysis (IVDA) Group at the University of Zurich is seeking a talented Ph.D. (m/f/d) at the intersection of visual analytics, interactive machine learning, and human-centered AI. Together, we will develop novel approaches for the characterization, design, and evaluation of interactive visual data analysis solutions that combine the strengths of both humans and algorithms. This position is particularly suitable for candidates with interests/experience in human preference elicitation and multi-criteria decision support, which we will study for the case of item ranking challenges.

Your responsibilities

Multi-criteria decision support through item ranking is an understudied research direction. Current solutions exhibit considerable limitations, primarily due to their inability to meet the multi-faceted nature of human decision-making. Choosing between thousands of relevant items is a common task, where people are left alone with the need to express and balance multiple desirable preferences at once. Item rankings are a popular and universal approach to structuring unorganized item collections by multiple criteria at the same time. In interactive solutions, people can express multiple preferences, used by algorithms to compute human-centered and personal rankings. Visual Analytics (VA) is a field of research that supports complex human decision-making tasks by bringing together human intellectuality with the computational power of algorithms in effective human-in-the-loop approaches.
In this project, we will demonstrate how cutting-edge VA principles can be transferred to item ranking, to overcome remaining challenges, demonstrate effective and efficient interactive ranking creation solutions, and empower broad audiences. Limitations of ranking systems mainly include their incompatibility for the elicitation of item- and attribute-based preferences, insufficient handling of user-introduced uncertainty, inadequate guidance for engaging with large item sets, lacking interaction histories, inefficient human-centric feedback loops, or a lack of explanations and transparency of ranking algorithms, thus creating a disconnect between the technology and nuanced user needs. Finally, interactive ranking creation does not yet make use of interfaces for eliciting implicit user preferences, e.g., in textual form using large language models.

Your profile

Requirements

• Master degree in computer science or a comparable subject from a recognized university
• Knowledge or interest in several of the following: visual analytics, information visualization, data science, information retrieval, data mining, machine/deep learning, AI, natural language processing, explainable AI, human-centered AI, responsible AI, algorithmic fairness, biases, transparency, HCI, applied research methods, or types of empirical research
• Profound skills in programming in Python
• Interest to work on applied research questions in a collaborative research environment

Desired Qualification and Skills

• Knowledge of user-centered design, design study methods, and evaluation methods
• Expertise with methods such as classification, regression, clustering, dimensionality reduction, similarity search, learn-to-rank, or recommender systems
• Experience in supervising students
• Ability and willingness to work effectively with students, faculty, and staff from all backgrounds

We offer:

• Possibility to achieve a PhD degree in computer science at UZH
• Hands-on supervision/mentorship for further career development

Support in conducting excellent research and in publishing results in top international journals and conferences

• Creative working atmosphere in a motivated, cooperative, and technically very well-equipped environment
• Possibility to work with several Ph.D. students in the lab in topics around IVDA, particularly on personalized VA for item ranking
• Excellent professional and personal development possibilities and hence, excellent basis for a future career in interactive visual data analysis in an industrial context
• Specific benefits like flexible working hours, young scientist promotion opportunities, parental leave benefits, nursery services, and care for dependents and much more
• Very good salary, according to local university regulations and standards in Switzerland

Information on your application

Applications must include a detailed CV, information on their educational background, professional references, a brief description of practical work and research experience, a clear exposition of prior data visualization experience, and a statement of motivation/expectation. Candidates are encouraged to visit the IVDA website.

Further information

Questions about the job

Prof. Jürgen Bernhard