The wait is over.

The Linguistic Fingerprint : A maths-only authorship identity verification method. A whitepaper created on Feb. 10, 2025, by Boban Dedović, Founder & Chief Psycholinguistics Officer, LINGA Co.

Abstract: In some cases, it is important to verify whether Person-A wrote Text-A. The Internet has made this task more difficult. Current technological solutions exist, albeit with challenges. Anti-plagiarism software generally relies on a large index of previously written documents in order to determine if a given text has similarities to anything else in the index. This approach helps us determine whether a given text is original, but it provides little value in determining whether Person-A wrote Text-A. Interested parties have tried to circumvent this drawback by means of requiring individuals to produce text in the company of others. The supervised setting is arguably robust because producing writing in person provides proof-of-work. However, the challenge noted above persists. Person-A can write Text-A, and Person-B can memorize the contents of Text-A in order to reproduce Text-A in a supervised setting. This issue of authorship identity verification remains. We therefore propose a novel method of identifying the author of a given text by means of mathematical operations—and these alone. For any given text, there are characters used to construct words, sentences, and paragraphs. The words are chosen, as well as their arrangements. For each character in a written text, we argue that there exist an almost infinite number of statistical relationships between each other character, and the nature of these patterns can be used to identify the true author of a given text. This is highly plausible because human beings learn language and grammar at a young age, and usually continue to refine the same throughout the lifespan. As a theoretical concept, this linguistic fingerprint should change over time based on the logical rules and constraints of a deterministic system. That is, the future state of a person‘s linguistic fingerprint necessarily relies on previous states, and there is no randomness to account for. This purely mathematical approach to identifying authorship based on statistical patterns in a given text seems both plausible and more robust than existing solutions. When taken to completion, the implications are legion.

By collecting individual datum points about writing patterns, we can normalize the values on a scale of 0–100 in order that we can visualize it.

Based on select factors, the radar chart below produces a geometric shape. This shape is unique to the individual, and changes over time according to other mathematical models we have developed. When we compare the shape of a given person‘s linguistic profile against another, we see that they are represented differently.

And by means of these mathematical operations, we can deliver statistical certainty as to whether Person-A actually wrote Text-A, and not a chatbot or someone else.