Analysis of existing natural language on semantic similarity

Ing. Yassine Ismaili

Master's thesis

Abstract:

The fourth industrial era is reliant on artificial intelligence (AI), and various semantic similarities approaches are constantly evolving. Theses popular techniques are widely applied in medical and health science, industrial and economic management. Similarity evaluation between textual documents is problematic and challenging. For this reason, the semantic text is used to provide a precise and structured representation of text meaning. It is largely used for solving text problems in various domains, as analyzing textual data, sentiment analysis, retrieving information, or extracting knowledge from textual data (text mining). For this purpose, the contribution of this research is to understand the meaning of semantic similarity among words and sentences and provide state of art, with their applications, to require more details about advantages and drawbacks. Also, in the second part, we aim to test the BERT methodology in the English and Czech languages, which is one of a powerful model for dealing with information retrieval and text summarization. This report describes the use of efficient and user-friendly semantic textual similarity tools to propose optimal techniques and solutions in terms of reducing time, resources, and computing costs.

Keywords

Embedding words, Semantic similarity, BERT, Deep learning

Language used: English

Date on which the thesis was submitted / produced: 3. 4. 2023

Thesis defence

• Date of defence: 22. 5. 2023
• Supervisor: doc. Ing. Arnošt Veselý, CSc.
• Reader: Richard Vágner, externi

Czech University of Life Sciences Prague

Faculty of Economics and Management

Master programme:
Informatics