Faithfulness Enforcement in Generative Multi-Text Summarization

Student: Ataev Aznaur

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Year of Graduation: 2024

Multidocument abstract summarization is the process of creating a summary that includes all or most of the essential information from a set of documents related by a common topic. Despite the progress made in this area, modern language models still face the problem of hallucinations — when a model generates text containing details or facts that do not correspond to known knowledge about the world or differ from the original input provided by the user. This problem has been widely studied in the field of text summarization, but much less research has been done on multidocument summarization. In this paper, we will discuss various techniques for improving the reliability of multidocument summarization, including mechanisms for increasing the plausibility of results by using information from related documents. We will mark our own dataset and look at the quality of the results obtained in comparison with the baseline.

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