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

Spell Checker for Bashkir Language

Student: Tatarinov Dmitriy

Supervisor: Boris Orekhov

Faculty: Faculty of Humanities

Educational Programme: Fundamental and Computational Linguistics (Bachelor)

Final Grade: 8

Year of Graduation: 2019

This research presents the development of an algorithm for automatic correction of printed texts in the Bashkir language. The purpose of this work is to study the existing solutions for creating spellcheckers, to describe the strengths and weaknesses of these solutions and develop its own algorithm, as well as to offer solutions to improve the quality of the developed model. As an example, the paper considers three common approaches to solving the problem of text classification and correction. First, these are models working with the help of neural networks. Second, models based on the methods of machine learning, using SVM models (Support Vector Machine) and models based on trees dependencies. Third, it is the construction of vector representations of text data with a further solution to the problem of separating the distance (difference) between the vectors constructed by their text representation. After studying the existing methods, it was decided to develop its own algorithm for checking printed texts based on the logic of hidden Markov models, as well as using the formula to calculate the probabilities of the existence of Markov chains. The main idea is that the sequence of characters in the language is natural and has its own mathematical dependence. Having calculated the probabilities of existence for all possible combinations, it is possible to analyze the new word form and determine the acceptability of this token in the paradigm of the Bashkir language on the basis of a sequence of symbols. In addition to the description of the development, there is also a description of the testing of the developed algorithm. Testing takes place in three stages using different types of data. The first test takes place on correct and specially generated incorrect word forms with mixed character order. The second test takes place on texts recognized using OCR technology. The final tests are based on articles written by people from Bashkir Wikipedia. As a result of the work, it was possible to develop a model for determining incorrect word forms in printed texts for the Bashkir language. The model has a flexible system of construction of processes that will allow to achieve the best result by means of preprocessing of texts or addition of more complex system of calculations.

Full text (added June 4, 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