Spot the Bot: Semantic Paths of Natural Language Texts

Student: Quynh nhu Dang

Educational Programme: Financial Technology and Data Analysis (Master)

Final Grade: 9

Year of Graduation: 2024

In the era of unlimited and potentially false information, it is crucial to identify which data has been artificially generated. With the rapid development of generative models, parallel training of detection models becomes an “arms race.” Therefore, it is important to examine the structural features of data generated by models of varying complexity. In this paper, we study persistent first-order homologies in the vector spaces of words, bigrams and trigrams of the Russian and English languages using topological data analysis. We select the most persistent homologies of largest diameter and obtain their contours. The features derived from these homologies make it possible to separate fiction texts from artificially generated texts.

Full text (added May 27, 2024)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses