Deep Learning for Sentiment Analysis of Short Texts

Student: Kovalev Evgeny

Educational Programme: Mathematics (Bachelor)

Final Grade: 10

Year of Graduation: 2018

Lately, deep learning techniques are widely used for solving problems of supervised learning. This is also true for sentiment analysis tasks. The goal of this research is to construct an architecture of the model for classification of short texts, namely, aggressive comments in the Internet based on the type of their toxicity. This work demonstrates that, in this task, deep learning models significantly outperform popular linear methods of machine learning. Also it is shown that due to the short average length of texts, such approaches as TF-IDF and convolutional layers prove ineffective. Finally, a combination of Bidirectional LSTM with Attention layer on top of it is proposed and implemented. The result of the model is 0.98 in AUC-ROC.

Full text (added May 31, 2018)

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