A key issue of semantics of information

In his epoch-making work on information theory, Shannon defended information in terms of entropy. Entropy-based definitions of information relate to quantity of information, but not to its meaning. Subsequent attempts to introduce semantics into information theory have made some progress but fell short of having a capability to deal with information described in natural language. This paper is aimed that of laying information for the theory which has this capability, call it a theory of semantics information (TSI). TSI is centered on a concept which plays a key role in human intelligence — A concept whose basic importance has long been and continues to be unrecognized — The concept of a restriction is pervasive in human cognition. Restrictions underlie the remarkable human ability to reason and make rational decisions in an environment of imprecision, uncertainty and incompleteness of information. Such environments are the norm in the real-world. Such environments have the traditional logical systems that become dysfunctional. There are many applications in which semantics of information plays an important role. Among such applications are: machine translation, summarization, search and decision-making under uncertainty. Informally, a restriction on a specified (focal) variable, X, written as R (X), is a statement which is a carrier of information about the values which X can take. Typically, restrictions are described in natural language. Example. X = length of time it takes to drive from Berkeley to SF Airport; R(X) = usually it takes about 90 minutes to drive from Berkeley to SF Airport. In adverse weather it may take close to 2 hours. An important issue in TSI is computation with restrictions. TSI opens the door to modes of computation in which approximation is accepted. Acceptance of approximate computations takes the calculus of restrictions (CR) into uncharted territory.