A User Study on Attribute Scoring Functions

Human-centered ranking of multi-dimensional items is a non-trivial task. Comparing complex items to each other in order to create a ranking can be time-consuming and especially if data sets are large, there is no guarantee that the result is satisfying. A shift from item granularity to attribute granularity might be beneficial to address complex decision-making problems. In this thesis, so-called Attribute Scoring Functions (ASFs) are studied which transform attribute values into numerical scores. We present a taxonomy of eight different types of ASFs based on identified ASFs in a literature review. ASFs can be used by users to express preferences for attribute values. The combination of multiple attribute scores allows the calculation of item summary scores that can then be used to create an item ranking. We propose a visual analytic approach that offers interfaces for all eight types of ASFs that supports a user in the creation of a ranking.

The example figure shows a case where a used car shall be bought, according to five criteria (attributes of cars). The charts show how the preferences of a user have been expressed an formalized using a visual interface. The result of the interactive tool yields top-ranked cars, to be approached.

We have created an interactive visual tool so that experts and non-experts can use ASFs more easily. To assess the usefulness of the tool, we want to do this evaluation. so we are searching for a student worker who helps conducting a user study in fall 2021.

Work tasks include:

• Experiment design and setup (together with the team)
• Experiment execution (as the "evaluator" of the participants of the study)
• Experiment data analysis (to identify findings of the experiments)

See the open positions page.