• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Text Punctuation Restoration

Student: Florinskiy Mikhail

Supervisor: Ekaterina Artemova

Faculty: Faculty of Computer Science

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 8

Year of Graduation: 2019

Text punctuation restoration is a difficult NLP task, which can be challenging even for a literate person. This implies that the source text has no punctuation marks. The solution to this problem can be applied in various spheres: automatic checking of punctuation in a text editing program, restoring punctuation marks and upper case registers in text obtained with the help of speech recognition, as well as correcting the result of the translator. Punctuation can be ambiguous and contextual. In addition, complex sentences can be split up into simpler ones, instead of using commas. The classical text punctuation approaches are based on parsing and pre-built structures using parts of speech and dictionaries. Despite high accuracy, they are incapable of coping with such cases. This problem can be partially or completely solved by LSTM neural networks. Moreover, such networks can provide a universal model that works with many languages ​​at the same time without being tied to any explicitly defined rules. The goal indicated by this study is to create an algorithm that restores punctuation in a coherent text in Russian, as well as to investigate how this algorithm would work with other languages. Bidirectional LSTM neural networks operating at the character level, CRF and a cross-lingual approach are used as methods for solving the tasks posed. The final model obtained by this study restores punctuation in text effectively.

Full text (added May 15, 2019)

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