Problem area 1: Developing simple methods for panel monitoring.
The focus here will be to identify a minimum number of simple tools and useful strategies for detecting the most frequently occurring noise in sensory data. It will here be important to evaluate and compare existing methods and also to find better and more explicit links between available methods and different individual differences that can occur. New method developments and/or modifications of existing methods may also be necessary in order to accommodate possible problems and shortcoming of existing methodology.
Problem area 2: Investigating and simplifying more advanced methods for panel monitoring.
The more technically advanced part of the project will be related to model-based methods taking all assessors and variables into account at the same time. These methods may represent an advantage due to the simultaneous utilisation of all available information, but the methods are rather complex and not so easy to use. To make these methods user-friendlier is an important challenge of interest for the whole community of sensory analysis. Pointing out advantages and disadvantages and clarifying what is gained by the extra efforts is also an important task that will be considered here.
Problem area 3: How to utilise the information detected during monitoring.
Here the focus is on developing and testing methods for combining information from a panel-monitoring process with the actual profile data themselves. The simplest strategy is to just delete the problematic data point, but this may sometimes lead to bias in the data and also far from optimal utilisation of information obtained. Certain types of weighting and other types of pre-processing may therefore be advantageous. Identifying the most efficient weighting scheme will be important here. Both the handling of individual attributes and the simultaneous handling of whole profiles will be important.
Problem area 4: Development of software for the methods developed.
The best from the method development above will be implemented in a software program. The program will be developed in close collaboration with the industries through frequent feedback concerning user-friendliness. The program will later be made available for other interested users via internet. The program will be supported and maintained by Matforsk, KVL and DTU.