The Half Past Human process for predicting the future from dynamic linguistics extracted from the internet began in 1997. Well, we have actually been gnawing on most of the concepts employed since about 1983. In 1994, we developed what we call the 'language model' for storage of data in large quanities for fast recovery when using SQL data bases. This extention of set theory as applied to data storage led to the examination of language in a new way. With the advent of the internet, the ability to test some of the 'language model' theories led to a series of programs being developed. This software is primarily written in prolog and implemented using LPA (Logic Programming Associates) version of this language as it offered pure prolog for the PC, as well as superior 'word concept' recognition and regex handling.

Our process begins with internet software agents which read in vast quantities of text from the public, and commonly accessible areas of the internet. We hunt for any of the words which are used as 'descriptors' within our process to define a context. Please see graphic below this discussion. These 'descriptors' are representative words or phrases which are used to define the basic 'idea' or 'concepts' of interest at that point.

Note we do NOT use 'conscious expressions' located on the internet, nor simple word counting techniques. We do not read emails. We only access publically posted texts on the internet seeking so much what is there, as what has recently changed there. So as a rule we concentrate our software agents on the forums and other community based sites.

What we do employ is a series of programming steps which reduce the text read down to 4/four digit hexidecimal integers which are themselves held in sets of 'inter linked linguistic discoveries'. These sets of hexidecimal integers are then aggregated along with information in a general sense as to where the text was located. The text returned is aggregated through further processing, producing a very large SQL based data base which is accessed by our prolog processing software. Once the data starts rolling in, the processing starts by associating the 'descriptor' with a whole group of 'values'. Together these form an 'aspect/attribute' coupling. In turn, the aspects/attributes are gathered into sets, and then humans examine these for various timing and manifestation clues.

The process of data gathering can take up to 3/three weeks to begin filling the data bases for processing. In total, in any given series , the data gathering will continue for an additional 3/three to 4/four weeks.

Once the interpretation is begun, a series of reports are prepared to present our findings in an entertaining and informative manner.