ONTOLOGICAL RESEARCH ARCHIVE INTEGRATION SYSTEM
Relevant for Research Area
Digital preservation in BrainLinks-BrainTools deals with the issue of how data produced during research is stored for future access and use. For this an archival system providing uniform descriptions and a web interface was developed during the initial project. To ensure that the archive is complete, it is crucial to provide elicitation tools that integrate into the work process. The first phase of this project will focus on analyzing existing workflows of the different research groups, identifying what documents need to be archived and how this can be achieved. Once the individual workflows have been analyzed, custom applications will be developed to fulfill the requirements. Archiving documents requires an individual elicitation process for each document type and type of workflow. For creating and optimizing these processes, current state of the art AI planning techniques will be applied using user preferences that are inferred from user interactions.
In our recent work we were able to show that preferences, so called soft goals, can be modeled using state dependent action costs and conditional effects. We proposed a method of improving the efficiency of planning in this setting by increasing heuristic accuracy. This can be achieved by a tighter integration of state dependent action costs and conditional effects within relaxation heuristics using a unified description of both based on edge-valued multi-valued decision diagrams.