Bachelor theses
Web application HashStory
Author
Alexandr Marinko
Year
2015
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
This thesis describes the development of web application HashStory. The main goal is to allow social network users to construct presentational timeline of story based on their own posts. The focus is put on social networks data access, asynchronous processing of import event and designing application programming interface (API) using architectural style REST (Representational state transfer). The result is a prototype of application that will serve for the future development.
Classifying Twitter election news as FakeNews
Author
Alina Vigriyanova
Year
2019
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
prof. Dr. Ing. Petr Kroha, CSc.
Department
Summary
This bachelor thesis aims to analyze Twitter archives of potentially state-backed Tweets, find a way of selecting reliable news from Twitter, obtain its counterpart of not Fake News, try different approaches to detect Fake News, analyze the approaches outcome, investigate possibilities of using topic modeling on the problem Fake News classification, and finally implement a tool that can help to classify FakeNews and evaluate the results. For text preprocessing NLP methods such as tokenization, stemming, stop words removal vectorizing were used. Logistic Regression, Linear Support Vector Classification, and Decision Tree classifiers were used to classify and evaluate the data. For topic modeling, Latent Dirichlet allocation was chosen.
Analysis tool for speeches of deputies
Author
Jan Horyna
Year
2020
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Lucie Svitáková
Department
Summary
Nowadays we live in the Information Age. Information is very valuable but there is so much of it therefore the risk of information overload does more often appear. In politics it is more complicated due to many political parties and each party has its own opinions. This thesis has an ambition to build a tool which purpose is to help citizens to orientate in speeches of members and parties in the Chamber of deputies Parliament of the Czech republic in the context of other speeches.
Three Java projects have been built in order to accomplish this task. The first has the purpose to download data from the parliament's website and prepare it for the second project which receives the data as an input. Subsequently it processes speeches and analyzes them in the tool MorphoDiTa. Finally the project transfers analyzed data together with other information and statistics to for this reason developed database. The third project performs information from the database to users. The presentation is implemented in the tool Vaadin which visualizes analyzed data in text and graph format. Members of parliament, parties or whole election periods can be compared to obtain improved context.
Architecture for big data stream processing
Author
Maroš Kramár
Year
2021
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Ivo Petr, Ph.D.
Department
Summary
In the last decade, data collection and processing is on the rise, thanks to expanding internet services and social media. In effort to speed up this process and obtain results in the shortest time possible, we meet with new challenges and problems compared to traditional batch approach. The thesis presents these issues and indicates solutions, that are used in modern open-source technologies used for data processing in real time. Based on presented technologies, a design of the environment, that allows execution of analytical applications is built. It is important that the whole system is reliable, scalable and can deliver precise results. This environment allows use of multiple frameworks that are compatible, used for implementation of analytic applications. Selected ones are compared from the standpoint of functionality and performance. Based on the design, a prototype is created, that allows running this environment on personal computer for testing purposes.
Scoring real estates based on customer personal preferences
Author
Roman Zhuravskyi
Year
2022
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Mgr. Eva Pernecká, Ph.D.
Department
Summary
In this thesis, we are going to get into the depths of the issue of residential real estate evaluation, study real estate portals to identify descriptive char- acteristics, and research factors influencing the decision of buyers. Also will explore appropriate sources of information in the form of open data or inter- faces of existing applications and will build a dataset including data related to property parameters and location. Research of the state-of-art articles about different recommendation engines was concluded, which resulted in the proposition of several variants of algorithms for property scoring based on available characteristics of the Czech real estate market. There were created drafts and tested two basic types of recommendation engines: collaborative filtering and content-based filtering. In the conclusion of this thesis, several recommendations are made and also proposed steps for future development.
Generování šachových pozic pomocí neuronových sítí
Author
Vladislav Bobko
Year
2022
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Mgr. Jan Spěvák, Ph.D.
Department
Summary
The bachelor thesis focuses on design and implementation of GAN architecture training framework for generating chess positions using generative neural networks. First, encodings for input chess positions are designed in order to transform them into a matrix form. Then basic types of neural networks are chosen for both generator and discriminator. Lastly, metrics are designed to compare the input chess positions with generated ones as well as the framework itself. Proposed models were trained on preprocessed dataset from server Lichess.org in Python language in interactive environment of Jupyter Notebook. In the end, experiments were performed and the quality of the proposed approach was evaluated. In the last experiment the trained model encapsulating proposed types of neural networks generates up to 29,7 % valid chess positions.
Fault detection in operations data
Author
Ostap Iudin
Year
2022
Type
Bachelor thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Petr Kasalický
Department
Summary
This bachelor's thesis is aimed at creating a full-fledged independent server monitoring system suitable for use in any company that owns any number of devices, and using machine learning to detect anomalies in the operation of technical devices to eliminate problems as soon as possible. As a result, I managed to create a monitoring system capable of monitoring the status of more than 6000 devices and an automatic anomaly detection system capable of predicting a failure of 6 hours before the failure itself.
Master theses
Tool for social analysis of forums
Author
Marek Horák
Year
2013
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Application for collecting social relations
Author
Jiří Mašek
Year
2012
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Michal Havryluk
Department
Monitoring tool of social networks
Author
Vojtěch Svoboda
Year
2012
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Tomáš Vitvar, Ph.D.
Department
Application to support teaching of Web services
Author
Tomáš Zeman
Year
2013
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Tomáš Vitvar, Ph.D.
Department
Modular web application as personal knowledge base
Author
Hynek Uhlíř
Year
2012
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Michal Havryluk
Department
Automation of interactions on Facebook
Author
Jakub Škvára
Year
2013
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Tomáš Vitvar, Ph.D.
Department
Information System To Support Community Organizing
Author
Slavomír Krivák
Year
2013
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
RNDr. Jakub Klímek, Ph.D.
Department
Caching REST server for cloud object stores
Author
Tomáš Plešek
Year
2013
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Mgr. Ladislava Smítková Janků, Ph.D.
Department
Module for graph clustering analysis
Author
Jiří Křížek
Year
2014
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
WebGephi - Web interface for Gephi
Author
Václav Čokrt
Year
2014
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
prof. RNDr. Tomáš Skopal, Ph.D.
Department
Testing of RESTful API in context of continuous integration
Author
Martin Štekl
Year
2014
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Marcel Hlopko
Department
Analysis of user behaviour in web applications
Author
Ondřej Musil
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Adaptable system for recommendation
Author
Jan Bouchner
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Ivo Lašek, Ph.D.
Department
Module for identification of key players in social networks
Author
Kristýna Streitová
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Tool for analysis of web technologies usage
Author
Martin Babický
Year
2015
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Implementation and integration of REST API to web portal focusing on a delivery service
Author
Jan Nedbal
Year
2015
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
The paper documents important decisions about developing RESTful API for a web portal focusing on delivery service. It provides solutions for some typical problems related to logistics and integration to existing system. Reader of this paper should be able to find here an information about overall design of API, used scaling, OAuth 2.0 authorization and Bearer authentication. The paper should provide complete overview of problematics related to creation of such system.
Web application for management of family trees
Author
Martina Plocová
Year
2016
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Jiří Mlejnek
Department
Summary
The thesis includes research of existing applications with focus on familytree creation and management. The analysis was conducted and according to analysis of requests from future application users. Based on the results from user research the web application for management of family trees was implemented. Application consists odf two parts, client and server. REST API enabling access to aplications data and other third-parties applications was designed and implemented as part of server side of application. After creating the application , the application was tested with selected users in which it was tested for functionality and usability of implemented application.
Tool for analysis of Czech news articles
Author
Markéta Filipová
Year
2016
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Tomáš Vitvar, Ph.D.
Department
Summary
Nowadays, when the amount of information on the internet continues to grow, automatic processing and analysis of data has become a very popular specialisation in the field of the information technologies. Online news service is one of the domains in which a significant amount of diverse as well as similar information exists. The goal of this thesis was to create a tool for analysis of Czech news articles which would simplify orientation in the data that is generated by these servers every day. To accomplish this goal a several Node.js modules have been created, each of which deals with a specific part of the process of obtaining interesting data. The first one is a crawler which allows downloading articles for analysis from news servers. In the second part, relevant content of articles and their other attributes are extracted from downloaded HTML pages. The third part is a text analysis for which modules for extraction of named entities and for sentiment analysis of Czech texts have been created. Finally, two CLI programs have been created, which allow easy download and analysis of articles, and the output of which are attributes of the article, its content, sentiment and a list of entities and its occurrences in the text. These articles have then been imported to the database Apache Solr for which several queries have been created and a number of experiments have been performed.
Learning bots for online games in a web browser
Author
Jakub Melezínek
Year
2017
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
RNDr. Jakub Klímek, Ph.D.
Department
Summary
Machine learning is used in many areas, including the Web. In the thesis it is used to implement self-learning bots that play web games right in a bowser. For decision making system I discuss options such as artificial neural networks or decision trees. At the end neuroevolution through augmenting topologies (NEAT) is chosen. As result of the thesis I created library written in TypeScript and Angular GUI application on top of it. It is possible to control evolution cycle through this application and see processed individuals. Two bots import the library. The first one plays game based on the pole-cart problem. The second one tries to play Slither.io, game similar to legendary Snake. Skripts are injected to game on client side and slightly changes the original game code. Results of the experiments suggest that application is working properly.
Tool for processing of EEG and its utilization in interactions with applications
Author
Filip Munzar
Year
2016
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Petr Pauš, Ph.D.
Department
Summary
This master thesis describes the design and implementation of library for cooperation with Emotiv EPOC neuroheadset and two applications that will use this library. Emotiv EPOC neuroheadset allows monitoring brain activity. The thesis gives a brief introduction to control a computer using human thoughts. There is described the Electroencephalography, the Brain-Computer Interface, a device used to connect the brain with a computer and finally there is the characteristic of Emotiv EPOC device. The core of this thesis is design, implementation and testing of library and application to recording data acquired usign Emotiv EPOC and tool for basic control of external applications with Emotic EPOC. The work is supplemented by installation and user manual.
Issue tracking system
Author
Lukáš Kořán
Year
2017
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Department
Summary
This diploma thesis deals with analysis, design and implementation of an issue tracking system. First part of the thesis describes analysis of existing solutions, ongoing processes while handling customer issues and requirements for a new system. The following part describes system design and the selection of suitable technology. System design consists of data modelling, system architecture and web application programming and user interface design. The last part summarizes development, testing, evaluates results of the development and possible further improvements. The result of system development consists of RESTful web service and web application. Web API is created using Spring platform, web client is developed with Angular framework. Web service is built on cloud service principles with user right management support. The issue management client application provides users with a user interface based on Material Design guidelines.
Ensuring scalability of web applications by using microservices architecture
Author
Lukáš Hamrla
Year
2017
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Tomáš Vondra, Ph.D.
Department
Summary
This thesis is focused on web applications scalability and high availability, especially for applications parts. Microservices architecture and container platform Docker are used to achieve this goal. There is a detailed analysis of these two technologies also. The thesis contains demo application which uses Docker and microservices and other supporting utilities. The text contains consequences, recommended procedures and possible problems and their solutions resulting from using these two technologies besides the application itself.
Application for real-time news recommendations
Author
Christián Golian
Year
2017
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
prof. Dr. Ing. Petr Kroha, CSc.
Department
Summary
This thesis deals with news recommendation and design and implementation of a recommender system.
The resulting program which took part in CLEF NewsREEL challenge uses association rules as a basis for recommendation.
The theoretical part of this thesis defines terms connected with recommendation systems and describes the specifics of news recommendation.
The practical part deals with design and implementation of recommender system rule-recommender.
Conclusions based on measurements made are drawn in the experimental part of this thesis.
Performance testing automatization of JavaScript environments
Author
Petr Kubín
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Pavel Štěpán
Department
Summary
There are many operating systems and browsers, enabling environment for running a program in JavaScript. The aim is to design a Web application, allowing you to select the environment (eg. operating system and Web browser) and test task (eg. specific JS code). Application will test the overall performance of the environment for the job (including operation count in specified amount of time).
Graph-based music player and manager
Author
Kryštof Šplíchal
Year
2017
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Josef Vogel, CSc.
Department
Summary
The goal of this thesis is to create an opensource web application for managing and playing a local music library with a new view on the data. The motivation is to allow the listeners of music to input custom relationships between the main entities. This results in a graph data structure. The application also follows the standards that are implemented in traditional music players. The graph structure is used for interactive visualisations, grouping of entities and generation of music playlists. The goal of those functions is to make browsing and using large music libraries easier for the end user. This has not yet been researched enough. The thesis describes the domain of music players, the process and results of analysis, design and implementation of the application and testing of the user interface.
I have created an application consisting of server and client parts. For the solution, I have used a graph database and various specific opensource libraries for the different parts. I have designed and implemented an algorithm for generation of music playlists based on the graph structure. I have created queries used for aggregation of track metadata from a linked data source. The testing has verified the functionality of implemented defined scenarios and has identified some problematic areas. The project is ready for further opensource community development.
The thesis gives a new method of working with a local music library and the created application allows the community to implement new features for better navigation in large music libraries using the application as the base.
Application for football league data collection and analysis
Author
Artyom Trushin
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
The main goal of this thesis is to implement a web application for gathering and analyzing football league statistics. The system will comprise of a web application and data providing server. The web application will be responsible for providing collected data and analysis to users. The data providing server will collect data and carry out analysis of it. Furthermore, a REST API for data transmission between two servers will be created.
Tool for detection and extraction of information from HTML
Author
Jakub Kottnauer
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
The bigger the Internet gets, the more important it is to be able to efficiently extract data from it. The main goal of this thesis is to implement a tool capable of automatic data extraction from web sites and to verify its capabilities in a set of experiments. The tool was successfully implemented in JavaScript and a series of experiments was run to verify that the tool works reliably when semantic annotations are present. The output of this thesis can be used either on its own or when developing other kinds of tools that work with the Web, such as crawlers or scrapers.
Framework for distributed computing on the web
Author
Jakub Šiller
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. David Bernhauer
Department
Summary
This master thesis focuses on design and implementation of a framework that uses computers of website visitors as computing nodes through web browsers. It contains an analysis of the Web environment, summarization of previous approaches and projects, design and implementation of the framework. The work describes the solution of computing node failure, reaction to slow computing node, possibilities of controlling the load of the framework on a website visitor's computer, options of asm.js and WebAssembly technology utilization, strategies for work distribution and security of the framework. At the end of the thesis, the experiment results and proposal of improvements are listed.
System for image search in augmented reality
Author
Petr Chmelař
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Department
Summary
The thesis deals with the problematics of computer vision, specifically analysis of image data, content-based image retrieval and possibilities for use in the augmented reality applications.
The main goal of this thesis is to implement a prototype of the system that will enable developers of mobile applications to define images through the web interface and search for them in the augmented reality on iOS platform. Individual chapters of the thesis deal with the analysis of requirements, the design of main parts of the system, the implementation itself and the subsequent testing.
The finished prototype of the system was deployed in the test mode and nothing prevents it's subsequent official release.
System for classification of web pages
Author
Stanislav Kusý
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Josef Pavlíček, Ph.D.
Department
Summary
This thesis describes automatic web page classification. Classification is composed
of three parts. The first part is downloading web page according to URL
address. The second part is extract and transform data from web page. The
third part is classification by machine learning algorithm. The result of this
thesis is application that assigns some of the predefined categories to the web
page.
Web application supporting semantic content management
Author
Michal Kopecký
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
This thesis deals with analysis of existing solutions for semantic data creation in content management systems. It also proposes its own solution that could make the process of creating structured data in CMS easier and more accessible to a wider group of users who do not have such an overview about semantic web.
These improvements are then implemented in the form of an appropriate extension of CMS Strapi. It adds a graphical tool for ontology modeling based on the Schema.org dictionary, and also allowing semantic data to create and publish.
In the final phase, this extension is subject of user testing, which verifies whether the result made semantic data creation easier and outlines its further possible development.
Implementation of anomaly detection in graphs
Author
Ondřej Filip
Year
2018
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Tomáš Zahradnický, Ph.D.
Department
Summary
Subject of this thesis is anomaly detection in graph. Thesis describes and eva-
luates several existing approaches dealing with this new and very little studied
problem domain. Two new and own approaches are presented, fist one exten-
ding one of existing algorithm and second one based of graph entropy changes
during its updates. One of the results is a library implementation containing
some selected existing methods along with two new one. All implemented
methods are evaluated and compared using synthetic and real-world graphs.
Web interface for the deployment and monitoring of Nomad jobs
Author
Pavel Peroutka
Year
2019
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Jan Ječmen
Department
Summary
This master thesis deals with the design and development process of an alternative user interface (UI) for HashiCorp Nomad. This UI is meant to be used by users without deeper understanding of IT operations. The resulting product called Bedouin is a fullstack web application, published as an open-source under the MIT license.
Data stream mining approaches in recommendation systems
Author
Tomáš Chládek
Year
2019
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Petr Špaček, Ph.D.
Department
Summary
Main topic of this master thesis is usage of data stream mining techniques in news recommendation systems. In theoretical part are described principles of recommendation systems, data mining and data streams. In previous work are revised existing algorithms in this domain and platform, that can be used for evaluation of recommendation system. In the data mining part is described the process of gathering information from the data stream and analyzed data, that are used for experiments. Streaming algorithm implements various parameterized techniques for data stream mining. Experiments are divided by following heuristics: random selection, popularity, iterator and recently visited article. Evaluation of experiments is performed on StreamingRec platform. The conclusion summarizes the benefits of using data stream mining techniques in the news recommendation systems.
Web application for annotations and recommendations of daily menus
Author
Michal Kváček
Year
2019
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
prof. Dr. Ing. Petr Kroha, CSc.
Department
Summary
This thesis describes analysis and implementation of simple recommender system used for recommending daily menus. Text covers the problem in complex way - from scraping data about daily menu to delivering final recommendation to users via web application. The final application is build as single page application on top of Vue.js framework. The recommender system was implemented as content based.
Algorithms for collaborative filtering in Point-of-Interest Recommendation Systems
Author
Guzel Samigullina
Year
2019
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
With the availability of the vast amount of users and Location-based social networks, the problem of POI recommendations has been widely studied and received significant research attention in the last seven years, and many approaches have been suggested. While previous works of POI recommendation mostly focused on investigating the spatial, temporal, and social influence, the use of additional content information has not been directionally studied. Such additional information can not only improve the performance of the recommendation system but also help to overcome the so-called cold start problem.
In this paper, we propose the content-aware matrix factorization method based on incorporating POI attribute and categories information. We propose two variants of the algorithm that can work with explicit and implicit feedback. The attribute and categories information of a POI is collected from existing datasets and used to measure the similarity between two POIs. These similarity values are subsequently used as a regularization term added to the objective function of matrix factorization. Experimental results show that the proposed method improves the quality of recommendation, outperforms most state-of-the-art collaborative filtering algorithms and can effectively cope with the so-called cold start problem.
Web application for recommendations of Points Of Interests
Author
Maryna Kryvosheienko
Year
2019
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
RNDr. Jakub Klímek, Ph.D.
Department
Summary
The main goal of this thesis is a design, implementation and evaluation of a proof of the concept web application acting as the Points Of Interests recommender system. The thesis focuses on the investigation of methods, that can be used for developing a recommender system, designing the hybrid algorithm, based on provided research, designing and implementing the web application which uses the proposed hybrid approach for making recommendations. Also, the thesis includes evaluation of the used method's results and results of the testing of the developed web application.
XML2API - universal REST API for XML documents
Author
Tomáš Špidlen
Year
2020
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Marek Sušický
Department
Summary
The goal of this thesis is to design and implement a software solution that allows access to static data in the form of XML files through a dynamic web API. The design and implementation took place on the basis of the requirements set by the contracting authority, the most important of which is the support of search based on the defined parameters. For the purposes of implementation, a group of current web technologies called MEAN stack was chosen. The result of the work is a functional prototype of a web application tested on a sample of real data provided by the client.
Open-source recommender system for Czech public procurement
Author
Milan Vancl
Year
2020
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Marek Sušický
Department
Summary
In this thesis I deal with the design and implementation of an open-source recommender system for Czech public procurement. I conduct research and discussion of current systems related to public procurrement. For the recommendation algorithm, I select a set of procurement features for which I design and implement their retrieval and extraction. I determine the main feature as the subject of procurement from which I extract semantic information using embedding. I demonstrate the recomender algorithm with a simple web application.
Application for processing videos to measure the speed of objects
Author
Daniel Šup
Year
2020
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. David Bernhauer
Department
Summary
The work focuses on the design of the library that detects objects in a video record, measures their speed and generates annotations, REST API and web applications for video processing and measuring the speed of objects in a video record. First, existing object detection methods, algorithms for measuring the speed of detected objects and video annotation methods are analyzed during the solution. Subsequently, suitable methods of object detection and suitable algorithms for measuring speed and methods of video annotation are selected. Then the library is designed and implemented in Python programming language. Furthermore, the REST API, which appropriately uses the functions of the library and ensures the detection of moving objects in the video, measuring their speed and generating annotations, is designed and implemented. Then the web application, which uses the REST API, is designed and implemented. In the work, the library that contains functions for detecting moving objects, measuring their speed and generating annotations, is implemented. Furthermore, the REST API, which uses the functions of the library, and the web application, which uses the implemented REST API, are implemented. The web application allows to upload a video, run a process for the video, which processes the video, and display the video with the calculated speeds processed by the process. The library and the REST API can be used as a basis of applications that do analysis of the objects based on their movement.
Machine learning and AI for web information extraction
Author
Martin Macho
Year
2022
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
This thesis deals with the topic of structured data extraction from websites. The aim of the thesis is to create a model for automated information extraction from the product pages of e-shops using machine learning and dataset created within this thesis. The theoretical part of the thesis deals with an introduction to the topic of automated data extraction, describes web mining, tools for information extraction from the website and methods of image processing using neural networks. The practical part focuses on the creation of the images of websites, the creation of a dataset and its use to train a model that identifies key elements on the product page and extracts them.
User interface and routing optimization in a courier company
Author
Jakub Lacný
Year
2020
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Vojtěch Jirkovský
Department
Summary
The purpose of this thesis is to analyse the internal processes in the MESSENGER courier company, and then to design, implement and evaluate the server part of a new optimization module. This module optimizes the process of assigning couriers to consignments. The results are a tested optimization module ready for gradual integration into the production environment, and the designs for the possible integration of the optimizations into the existing user interface. The current real-time load of couriers and analysis of pickup and delivery places are used to optimize assigning. The module is implemented using Java programming language with the help of the Spring and Hibernate frameworks. The data is stored and queried with MSSQL and ElasticSearch.
Processing streams of short texts in real time
Author
Jakub Hynek
Year
2021
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
prof. Ing. RNDr. Martin Holeňa, CSc.
Department
Summary
Topic of this thesis is processing of on-line live streams comments to enable better interaction. The focus is to do a research of existing stream clustering algorithms and experimentally try them on real data. In the thesis is a research of existing algorithms and implementations and a solution, where text is preprocessed in programming language python with the use of libraries NLTK and Gensim and clustering performed in java framework for stream learning MOA.
Document management web-app for a small group
Author
Daniel Šulik
Year
2021
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Jiří Daněček
Department
Summary
The thesis deals with designing and implementation of electronic document management system intended for individuals or a small group of people. The application itself is divided into four parts: Front-end, Back-end, OCR-API and database. Front-end part is implemented in Angular 10. Back-end and OCR-API are coded in Java with exposed REST API endpoints. The database is using MongoDB technology. Lastly, every of the mentioned components is made into a Docker image available to use.
Search optimization in the case-law of supreme courts of the Czech Republic
Author
David Záleský
Year
2021
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
This thesis focuses on the effectiveness of searching the databases of decisions of the supreme courts of the Czech Republic. It describes existing solutions and their shortcomings, and outlines the ways how to make the significant decisions in individual legal matters more easily accessible to both the courts themselves and the general public. The selected solutions are then implemented in the Python programming language.
Crossword Generator using Web Data
Author
Adam Benda
Year
2021
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Ivan Šimeček, Ph.D.
Department
Summary
This Thesis develops a Crossword Generation software, sourcing the list of words and semantic word's categorizations from Wikidata, the Linked Data Knowledge Base. A crossword constructing heuristic is implemented and modified to prefer user-defined categories of words. A simple web-based user interface is developed and the project parts are interconnected to form a full stack automated crossword generator.
Server-side application for weekly meal planner
Author
Pavel Baroš
Year
2022
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
The thesis describes the design and implementation of a server service for planning and recommending a meal menu from an available recipe database. The thesis describes all phases of the development of such a system - from the definition of business requirements, acquaintance with the theory of recommendation systems, to the evaluation of selected algorithms. The web service is implemented as a RESTful API. The system is hybrid, combining popular item recommendation and content-based filtering.
Extraction of company descriptors from web resources
Author
Tomáš Stanovčák
Year
2022
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Pavel Kordík, Ph.D.
Department
Summary
The subject of this thesis is to obtain and process company data from their websites. After getting acquainted with extraction approaches and available set of company information, dataset will be prepared in a format suitable for experiments. This dataset will undergo the extraction procedures based on both rule and machine learning principles. The results of the experiments will be evaluated and the implementation of the individual approaches will be publicly accessible as a library under a free licence.
Server-side application for CAPTCHA system
Author
Otakar Vinklář
Year
2022
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
Today one of the ways how to secure applications against malicious bots is
to use CAPTCHA (Completely Automated Public Turing test to tell Com-
puters and Humans Apart). But usage of CAPTCHA might have issues.
For some users CAPTCHA might be hard to solve and also take some of
their time. We have implemented a CAPTCHA application, where website
providers can configure the CAPTCHA verification method in a way that it
suites their users and is as comfortable as possible. Next the website providers
are able to use the valuable human computing power, which is harnessed from
the tested users. Further the CAPTCHA application allows programmers
to simply add new verification methods. We have implemented two default
verification methods which demonstrate the described application properties.
One of the verification methods is image-based selection CAPTCHA, which
allows website providers to specify images used for the task generation and
also enables them to specify, which labels should appear in the task question.
Furthermore, website providers can add their own images to shown to their
users. In addition this CAPTCHA utilizes human computing, so the users can
add unlabeled images, which are then labeled
Fraudulent behavior detection using graph-based anomaly detection approaches
Author
Ondřej Šofr
Year
2023
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
This thesis deals with the field of anomaly detection. Traditional and modern approaches are described and analyzed, especially the methods of graph-based anomaly detection and their ability to utilize the relational information in the data. The main focus is the processing and modeling of user actions in order to detect their fraudulent behavior in several domains. Prediction accuracy and efficiency of solutions are evaluated with the emphasis on the usability in practice. This thesis contains experimental results of presented methods tested on real-world data containing posts extracted from social networks. The task is to detect misinformation claims using the structure of related discussion threads. It is shown that there is a connection between anomaly detection and misinformation detection and that it is possible to achieve accurate results using these approaches.
A tool for changing and simulating the user context on the internet
Author
Marek Király
Year
2023
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. Jiří Kříž
Department
Summary
The thesis aimed to design and implement a tool that allows changing the browsing context, which carries within data that allow determining specific information about users on the Internet. The tool is implemented in the form of an extension designed for the Google Chrome browser, and in the design it was necessary to take into account the technical possibilities and barriers that are intended for browser extensions. After implementation, user testing was conducted which showed that the application works properly and is usable while browsing internet content.
Cloud bot for game Metin2
Author
Daniel Jankech
Year
2023
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
doc. Ing. Tomáš Vitvar, Ph.D.
Department
Summary
This master thesis proposes the development of a cloud-based bot for the massively multiplayer online role-playing game (MMORPG) Metin2, which will be designed to perform tasks such as farming for in-game resources and completing quests. The bot will be hosted in the cloud, providing horizontal scalability and reliability. However, developing a cloud-based bot for Metin2 poses several challenges, including designing the bot to interact with the game's client undetectably, try to navigate the game's environment indistinguishably from human behavior, and ensure scalability and reliability in a cloud-based environment. This thesis explores the technologies and techniques that can be used to develop such a bot and proposes solutions to overcome these challenges. The proposed cloud-based bot for Metin2 aims to improve the gaming experience for players by enabling them to progress more efficiently through the game.
Recommender System for a Bank Application
Author
Miloš Popovič
Year
2024
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
PhDr. Ing. Tomáš Evan, Ph.D.
Department
Summary
The goal of this work is to design and implement a recommender system in the form of a microservice aimed at corporate banking clients. The theoretical part analyzes the banking domain, data and processes, conducts research on collaborative filtering recommendation systems, and describes the architecture of microservices. In the practical part, experiments are conducted with selected algorithms, the architecture of microservices for an application for designing and recommending products is proposed, and a recommendation microservice is implemented. The result of the experiments is a hybrid algorithm in the form of clustering applied to clients' analytical data and the application of similarity functions of user-based filtering to product and fee data. Furthermore, five new microservices were designed and successfully integrated among other banking systems. Finally, a microservice for recommendation was successfully implemented using an asynchronous principle, a message broker and a Python web application framework.
Integration platform for LMS
Author
Martin Suchan
Year
2024
Type
Master thesis
Supervisor
Ing. Jaroslav Kuchař, Ph.D.
Reviewers
Ing. David Buchtela, Ph.D.
Department
Summary
Objective of this thesis is to create integration platform for the unification of various types of integrations on an already existing learning management system. Integration can manage any administrator without knowledge of any programming language. Integrations are written into JSON file called metadata, where data format and method of obtaining data of the integration is described. Integration platform can obtain data via local server, remote server (SFTP, FTPS), remote server with web service or can provide own web service server. The thesis contains the design and implementation of the integration platform, as well as several examples of desribing integration into a metadata file.