Rhetorical relations include CAUSE_MARKERS = { " because ", " since ", " as a result of ", " due to ", " owing to ", " in consequence of ", }, RESULT_MARKERS = {" therefore ", " thus ", " hence ", " consequently ", " as a result ", " so "}CONTRAST_MARKERS = { " but ", " however ", " yet ", " whereas ", " while ", " in contrast ", " on the other hand ", (...) }, ONCESSION_MARKERS = { " although ", " though ", " even though ", " despite ", " in spite of ", " nevertheless ", " nonetheless ", " still "}, CONDITION_MARKERS = {" if ", " unless ", " provided that ", " in case ", " as long as "},EXAMPLE_MARKERS = {" for example ", " for instance ", " such as ", " e.g. ", " e.g. "}RESTATEMENT_MARKERS = {" in other words ", " that is ", " i.e. ", " namely "}, TEMPORAL_MARKERS = { " then ", " afterwards ", " afterward ", " before ", " after ", " subsequently ", " meanwhile ",} and ATTRIBUTION_MARKERS = { " according to ", " reported ", " reports ", " said ", " claims ", " claimed ", " stated ". Rhetoricality is a random variable defined on the space of intervals of sentences in a (PDF) document. It is the count of the total number of directed rhetorical relations between the sentences in the interval and all other sentences in the document. Rhetoricality is considered to be closely associated with semantics, the intended meaning of the document. Prevalence is also a random variable defined on that space of sentences. It is a calculation of the density of high-frequency words in an interval of sentences. Its virtue is that it is much more efficiently calculated than rhetoricality. The driving conjecture of this research is that Prevalence and Rhetoricality are significantly correlated in the sense of statistics. Should this conjecture be born out, it augures computationally rapid comparison of meaning among documents. This document is a compendium of computed evidence in support of the conjecture. Since the author may be getting ahead of himself, and the large language model ChatGPT (affectionately called, “gipity”) used to assist coding is entirely capable of subtle hallucinations, this work may be merely in support of a delusion born of a human and a computer. Therefore, the code and database are available on Github, and the hope is that the results are replicable by sufficiently experienced Python programmers. If so, then the correlation of prevalence and rhetoricality may be a scientific discovery. (shrink)
