Authors:

Ashwin Sreedhar

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

This paper proposes a novel encryption method leveraging semantic word embeddings generated via sense2vec for encoding plaintext messages into coherent encrypted text. In today's digital age, encryption is necessary for securing sensitive text data. Traditional encryption methods render ciphertext as seemingly random sequences of characters, preventing any readability or contextual understanding without decryption. However, our approach maintains a degree of semantic coherence in the ciphertext, making it appear as comprehensible text while preserving security. This research outlines an encryption algorithm that applies spectral clustering to sense2vec-generated embeddings; these clusters are then mapped to plaintext words. By selecting random words within the clusters, a ciphertext is generated. This ensures that the resulting ciphertext remains coherent, providing a semblance of natural language text. By masking the true content behind contextually similar but inherently distinct wording, the methodology protects the original message while obfuscating its meaning to unauthorized readers. The paper further assesses the semantic coherence of the encrypted text through verifying that natural language processing tasks such as authorship attribution and text classification are capable on the ciphertext. The results provide strong evidence for the robustness and effectiveness of this encryption method, demonstrating its potential for securing sensitive textual data while evading censorship or detection efforts that target traditional encryption artifacts. By advancing the intersection of cryptography and natural language processing, this paper introduces a new paradigm in text encryption, highlighting the potential for further research in this field.

Source:

Purdue University / 2023

Topics:

No topics listed

Co-authors:

Ashwin Sreedhar