Paper Summary - An Introduction to Multisensor Data Fusion

D. L. Hall and J. Llinas, editors. Handbook of Multisensor
Data Fusion. CRC Press, 2001.

This paper is based on the book and gives a general background in multisensor data fusion. It gives basic definitions, a bit of history and highlights types of applications that use multisensor data fusion techniques. This is most prevalent in military applications but commercial applications are also making use of fusing data from multiple sources. There are advantages in using a 'multi-sensor' approach, improved accuracy and estimates are better and in general there is a statistical advantage.

It goes on to provide some basic definitions and discusses examples of sensors (more related to military domain). What is interesting is the following:

"The most fundamental characterization of data fusion involves a hierarchical transformation between observed energy or parameters (provided by multiple sources as input) and a decision or inference (produced by fusion estimation and/or inference processes) regarding the location, characteristics, and identity of an entity, and an interpretation of the observed entity in the context of a
surrounding environment and relationships to other entities....
The transformation between observed energy or parameters and a decision or
inference proceeds from an observed signal to progressively more abstract concepts."

They go on to discuss methods which are to make identity estimations including Dempster-Shafer and Bayesian.

"Observational data may be combined, or fused, at a variety of levels from the raw data (or observation) level to a state vector level, or at the decision level."

It then talks in details about examples in military and non-military applications and then about the Joint Data Fusion Process Model which was established in 1986.

The rest of the paper goes into detail about the architecture.

Why is this important to my work?

There are aspects about true multi-sensor data fusion that can be adapted and used in fusing semantic web data. There are very similar issues involved. We get data about entities from different sources. This data can be complementary, certain sources can offer facts that other sources are not aware of and fusing this information together presents a more comprehensive picture of an entity. This is applicable which smushing FOAF instances (part of earlier work) and this is applicable which simply merging fact retrieved from different sources. One example in particular is new sources. Different facts can be exposed from different news sources. When you bring these facts together you get a more complete story.

This brings us to another issue with data fusion and that is conflict resolution. When we are combining sources sometimes the information can be in conflict with each other. This is an interesting problem.

This paper is a great way to get a good background in multi-sensor data fusion and one can use definitions, techniques, architectures and apply them to fusing Semantic Web data.