Relations among science indicators or more generally among anything one might wish to count about texts

It is shown that the expected information content of distributions provides us with a straightforward means to develop a static and a dynamic model for the development of the sciences.

Abstract

In a series of two articles, I will show that the expected information content of distributions provides us with a straightforward means to develop a static and a dynamic model for the development of the sciences. In the first study, I analyze how knowledge about one indicator (nominal variable) can reduce our uncertainty in the prediction of other indicators, and how relations across various levels of aggregation can be assessed. In the second study, I will address the problem of the use of indicators and relations among them for predictions and reconstructions. I will use the occurrences of words in texts as the prime nominal variable which can be easily counted by the machine. However, I will generalize the models for the multi-variate case, in which any indicator or nominal variable can be assessed in terms of its validity in relation to other indicators and its value for predictions.