New Interpretation of the Shannon Entropy Measure.

摘要

Although more than sixty years have elapsed since Shannon's seminal information entropy paper the literature reveals that there are divergent opinions of what it actually measures. From its similarity to Boltzman entropy in statistical mechanics, the most common view is that it measures information disorder and uncertainty. Based on an inductive derivation of the expression we propose a new interpretation relating to the use of symbols to uniquely identify probabilistic messages. Applying this new meaning to Shannon's English language experiment we extract a new interpretation of those results. Moreover, the new understanding of Shannon entropy also has the potential to improve the effectiveness of intelligence analysis applications.