Semi-automatic assessment of User Experience based on a battery of machine learning and artificial intelligence methods using information from various sources (e.g. interaction data, web-services, usability heuristics)


Motivation & Project Description

• UX has become the key success metric for all commercial services and products
• UX is on the DTAG roadmap of most urgent topics
• Fusion of several engineering approaches for a quick and cheap objective assessment of a design's UX
• The proposed methods are tested with several design prototypes

Outcome:

• Model of the user’s belief about the dialog state; can be used to predict quality judgments and detect problematic dialog situations
• Activation-based user model – first step towards a model of users’ affect during the interaction
• Analysis of importance of quality aspects on overall quality judgments of dialog systems
• Interaction quantification framework and logging capabilities for android apps
• Model of older adults’ errors when using smartphones due to their mental model of the device
• Models to predict modality selection in multimodal interaction
• Models to predict perceived quality in multimodal interaction
• Study on user characteristics and user judgments for android apps (still running)
• Study on app characteristics and user judgments for android apps (still running)
• Several papers and a framework for research on dialog simulation
• Improvements in the MeMo workbench

Time Frame: 01/2011 - 12/2012

T-labs Team Members: F. Gödde, K.-P. Engelbrecht, S. Möller, S.Schmidt, M. Schulz, B. Weiss

Students: M. Siebke, R. Tönges, G. Neitzel

Funding by: Deutsche Telekom Laboratories