Generalised structures for the composition of evidenceFunded under: JRC-ADVMAT 1C
The problem of defining flexible generalised rules to combine bodies of knowledge is considered in the frame of evidence theory. The starting point of the analysis is the non-normalised Dempster rule of combination and its interpretation in terms of a two level process of merging and projection. It is argued that triangular norms and conorms can provide a consistent and coherent framework. The generalised composition rule which ensues can be specified by ascribing a proper semantics to each composition pattern. As a result, it becomes possible to handle the analogy with the common geometric operations, as well as the properties of the basic probability assignments. The generalised composition rule is dealt with as an example of a quasi-associative operator. Thus the process can be interpreted as a step in the progressive accumulation of evidence. In turn, the accumulation of evidence originates a process of learning which can be linked to the absence of associative property.
Bibliographic Reference: Paper presented: Third IFSA Congress, Seattle (US), Aug. 6-11, 1989
Availability: Available from (1) as Paper EN 35041 ORA
Record Number: 198911222 / Last updated on: 1994-12-01
Original language: en
Available languages: en