Quoting Loet Leydesdorff <[email protected]>:

>
> Because it is essentially a mathematical theory, information theory,  
> in my opinion, can be made relevant (that is, provided with meaning)  
> in all kind of contexts, and these various contexts can be used to  
> formalize mechanisms which can then be more abstractly translated  
> into other contexts by using information theory.

Dear Loet,

Your opinion that IT can address meaning is refreshing, and I quite  
agree with you. Most feel that the Shannon approach cannot address  
meaning. I offer, however, the following example:


Consider the following three random strings of 200 digits:

Sequence A:

42607951361038986072061245134123237671363847519601557824849686201007746224524209371591449046940565604803389860720612451341232376713638475196015578248496862010077462245242093715914490469405656048033898

Sequence B:

03617746439242093715914490469405656048033898607206124513412323767136384751960155782484968620100774622452420937159144904694056560480338986072061245134123237671363847519601557824849686201007746224524209

Sequence C:

01475623843789694751743102380318185453848905236473225910906494173735504160210176853263006704607242470971896947517431023803181854538489052364732259109064941737795041102101768532630067046072424709708969

The values of the Shannon measure for each sequence are 3.298, 3.288  
and 3.296 bits, respectively. That there exists no internal order in  
any of the sequences is shown by calculating the mutual information  
measures of adjacent pairs of digits in each of the three cases. These  
amount to 10.97%, 10.03% and 9.94% of the respective paired entropies,  
each of which is typical of a random distribution of these dimensions.  
Relationships between higher lagged pairs prove likewise to be random.

Next the correspondences between pairs of sequences are examined.  
Looking at how each digit in A pairs with its corresponding location  
in B yields a joint entropy of 5.900 bits, 11.61% of which appears as  
mutual information – i.e., random correspondence. Similar pairings  
between sequences A and C, however, reveal that fully 91.69% of the  
joint entropy consists of mutual information between the sequences.  
Obviously, sequences A and C are closely related. In fact, close  
scrutiny of sequence C shows it to be an arbitrary permutation of  
sequence A, along with a handful of “mistaken” permutations.

While this may appear as a typical exercise in coding/decoding, its  
implications actually run far deeper. Instead of digits, one could  
have used symbols for codons in a genome (A,C,T,G) or monomers in a  
protein. In the latter instance, sequence A might represent the  
spatial sequencing on the outer surface of an antibody in the plasma  
of an organism, and B and C might describe the patterns on the  
surfaces of microbes present in the same plasma. While B appears to  
bear no relationship to A, C would engage A in a “hand- in- glove”  
match.

The pattern in C would have ultimate meaning to A. It would signify  
the goal towards which A was created by the immune system, and the  
match would initiate a highly directed action (to eliminate the  
microbe). All of this significance can be gauged by the elevated value  
of mutual information between the sequences. So, whereas the raw  
Shannon measure by itself cannot convey meaning, it is obvious that  
the relative information conveyed by the AMI is capable of doing so.  
That “meaning” for antibodies is a pale representation of meaning in  
the human context only reflects how wanly quantitative models in  
general prefigure more complicated human situations. The basic  
elements of meaning, however, can indeed be captured by relativistic  
information measures.

Greetings to all,
Bob


_______________________________________________
fis mailing list
[email protected]
https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis

Reply via email to