Context:
Text filtering is often used for finding suitable data for given task and improving model quality.
There are numerous tasks benefiting from filtered text data such as spam detection, LLM assistent creation and fine-tuning, or Machine translation.
Since there are many text filters, it might be impossible to exhaustively evaluate all of them, however, it is worth doing an overview on few such methods in the field of machine translation and creating a way to compare them.

Goals:
The main goal of this thesis is to explore and implement multiple sentence filtering approaches, and test them using machine translation measures using Czech-English data. An important part of the thesis shall be an evaluation the of filters and a way to evaluate a given filter on machine translation.

Specifically, the filters will be used to create data from which machine translation models will be trained and tested using measures such as ChrF or Comet.
A large part of the focus of the thesis shall be on the evaluation/comparison
of filtering methods on machine translation to help compare different filtering methods.

Requirements:
The final work shall consist of:
- Comparing few modern approaches of text filtering methods for machine translation.
- Implementing/importing some such approaches and describing them in the thesis. Two of these methods shall be Fastext based filltering and embedding+neural network classifier.
- Evaluating the filtering approaches using machine translation measures.
- A written thesis containing results of the work done