黑料不打烊

艩iame bakalauro darbe 鈥濼uristini懦 kelioni懦 rekomendacin臈 sistema鈥� analizuojami teoriniai ir praktiniai sprendimai, susij臋 su turistini懦 lankytin懦 viet懦 rekomendavimu mobiliosios 鈥濧ndroid鈥� operacin臈s sistemos programoje. Sukurta mobilioji programa skirta personalizuot懦 kelioni懦 pasi奴lym懦 k奴rimui Lietuvos teritorijoje. Programos k奴rimui naudota 鈥濧ndroid Studio鈥� aplinka su 鈥濳otlin鈥� programavimo kalba. Lokaliam duomen懦 saugojimui pasirinkta 鈥濻QLite鈥� duomen懦 bazi懦 valdymo sistema. Darbas suskirstytas 寞 keturias dalis: analiz臈s, projektavimo, realizacijos ir testavimo. Analiz臈s dalyje ap啪velgiami aktual奴s literat奴ros 拧altiniai, rinkoje esamos turistini懦 kelioni懦 rekomendacin臈s mobiliosios programos ir mobili懦j懦 program懦 k奴rimo technologijos. Projektavimo dalyje pateikiami funkciniai ir nefunkciniai reikalavimai, naudojimo scenarijai, proces懦 eiga ir strukt奴rinis modelis. Realizacijos dalyje apra拧omos naudotos technologijos ir 寞rankiai, pateikiami sukurti vartotojo s膮sajos langai bei 寞gyvendintos funkcijos. Testavimo dalyje 寞vertinama programos kokyb臈, atliekant vienet懦 ir rankinius testavimus, siekiant patvirtinti atitikim膮 projektavimo dalyje apibr臈啪tiems reikalavimams. Darbo pabaigoje pateikiamos apibendrinan膷ios i拧vados, apiman膷ios vis懦 dali懦 rezultatus ir 寞啪valgas. Strukt奴ra: 寞vadas, analiz臈s dalis, projektavimo dalis, realizacijos dalis, testavimo dalis, i拧vados ir literat奴ros s膮ra拧as. Darb膮 sudaro: 91 puslapis, 9 lentel臈s, 56 iliustracijos ir 37 literat奴riniai 拧altiniai.

This bachelor's thesis "Touristic trip recommendation system" analyses the theoretical and practical solutions related to the recommendation of tourist attractions in a mobile application for the "Android" operating system. The developed mobile application is designed for the creation of personalized tourist attractions in the territory of Lithuania. The "Android Studio" environment with "Kotlin" programming language was used to develop the application. The "SQLite" database management system was chosen for local data storage. The work is divided into four parts: analysis, design, implementation and testing. The analysis part reviews relevant literature sources, the tourist travel recommendation mobile apps available on the market and mobile app development technologies. The design part presents functional and non-functional requirements, usage scenarios, process flow and structural model. The implementation part describes the technologies and tools used, the user interface windows developed and the functions implemented. The testing part assesses the quality of the application by performing unit and manual tests to validate compliance with the requirements defined in the design part. The thesis concludes with a summary of the results and insights from all parts. Structure: introduction, analysis section, design section, implementation section, testing section, conclusions and reference list. Thesis consists of: 91 pages, 9 tables, 56 illustrations and 37 references.

Baigiamojo bakalauro darbo tikslas 鈥� palengvinti augintini懦 savininkams prie啪i奴ros paslaug懦 u啪sakymo ir valdymo proces膮, sukuriant mobili膮j膮 program臈l臋. Analiz臈s dalyje i拧analizuotas augintini懦 prie啪i奴ros paslaug懦 sistem懦 poreikis, nustatyti svarbiausi funkciniai ir nefunkciniai reikalavimai bei aptarti esam懦 sprendim懦 privalumai ir tr奴kumai. Remiantis i拧keltais reikalavimais, atlikta sistemos u啪duo膷i懦, proces懦 bei duomen懦 modeli懦 detalizacija, atsi啪velgiant 寞 vartotoj懦 poreikius. Remiantis reikalavim懦 specifikacija, sukurta mobiliosios 辫谤辞驳谤补尘臈濒臈s sistema ir atliktas jos testavimas, 寞skaitant funkcionalumo patikrinim膮 rankiniais ir automatizuotais metodais. Testavimo rezultatai patvirtino sistemos atitikt寞 keltiems reikalavimams. Darbo strukt奴ra: 寞vadas, augintini懦 prie啪i奴ros paslaug懦 analiz臈, sistemos projektavimas, mobilios 辫谤辞驳谤补尘臈濒臈s realizacija ir testavimas, i拧vados. Darbo apimtis: 55 puslapiai, 34 lentel臈s ir 25 paveiksl臈liai. Naudota 32 literat奴ros 拧altiniai.

The aim of this bachelor鈥檚 thesis is to facilitate the process of ordering and managing pet care services for pet owners by developing a mobile application. The analysis section examines the need for pet care service systems, identifies key functional and non-functional requirements, and discusses the advantages and shortcomings of existing solutions. Based on the identified requirements, detailed modeling of system tasks, processes, and data was performed, taking into account user needs. According to the requirements specification, the mobile application system was developed and tested, including functionality verification through manual and automated methods. Testing results confirmed the system鈥檚 compliance with the specified requirements. The thesis structure includes an introduction, pet care services analysis, system design, mobile application implementation and testing, and conclusions. The thesis comprises 55 pages, 34 tables, and 25 figures. A total of 32 literature sources were used.

Baigiamajame bakalauro darbe sukurtas edukacinio nuotyki懦 啪aidimo prototipas, skirtas moksleivi懦 programavimo 啪ini懦 ugdymui. Darbo tikslas 鈥� i拧analizuoti 啪aidybinimo metod懦 panaudojimo galimybes mokant programavimo pagrind懦 bei sukurti interaktyv懦 edukacin寞 啪aidim膮, kuris pad臈t懦 efektyviau 寞sisavinti teorines 啪inias per praktin寞 patyrim膮. Darbo analitin臈je dalyje i拧nagrin臈tos programavimo mokymo problemos, 啪aidybinimo principai bei pana拧奴s rinkoje esantys edukaciniai 啪aidimai, atlikta technologij懦 ap啪valga. Projektin臈je dalyje buvo sukurti funkciniai ir nefunkciniai reikalavimai, panaudos atvej懦, sek懦, klasi懦 ir komponent懦 diagramos, suprojektuota 啪aidimo architekt奴ra. Realizacija atlikta naudojant Unity 啪aidimo varikl寞, C# programavimo kalb膮 ir integruot膮 k奴rimo aplink膮 Visual Studio. Sukurtas 啪aidimo prototipas buvo i拧testuotas rankiniu b奴du bei naudojant specialiai parengtus scenarijus. Praktinis 啪aidimo testavimas mokykloje parod臈, kad 啪aidimas yra 寞traukiantis, veiksmingas ir tinkamas 寞vairi懦 啪ini懦 lygi懦 moksleiviams. Darbo apimtis 鈥� 67 puslapiai be pried懦, 13 lenteli懦, 52 iliustracijos. Panaudota 24 literat奴ros 拧altiniai.

In the final work of the bachelor thesis a prototype of an educational adventure game was developed for the development of students' programming knowledge. The aim of the thesis is to analyse the possibilities of using gamification methods to teach programming basics and to create an interactive educational game that would help to acquire theoretical knowledge more efficiently through hands-on experience. The analytical part of the thesis analyses the problems of teaching programming, the principles of gamification and similar educational games available on the market, as well as an overview of technologies. In the design part, functional and non-functional requirements, use case, sequence, class and component diagrams were developed, and the game architecture was designed. The implementation was carried out using the Unity game engine, the C# programming language and the integrated development environment Visual Studio. The prototype game was tested manually and with specially designed scenarios. Practical testing of the game in a school has shown that the game is engaging, effective and suitable for students of different knowledge levels. The thesis consists of 67 pages without appendices, 13 tables and 52 illustrations. 24 references used.

Baigiamojo bakalauro darbo pagrindinis tikslas 鈥� palengvinti kelioni懦 motociklu planavim膮, sukuriant mobili膮j膮 program臈l臋, kuri automati拧kai generuoja mar拧rutus pagal pasirinktus kriterijus. Analitin臈je dalyje atlikta i拧sami kelioni懦 planavimo ir sekimo program臈li懦 analiz臈, nustatyti pagrindiniai funkcionalumai ir reikalavimai. Projektin臈je dalyje nustatyti funkciniai ir nefunkciniai reikalavimai, atlikta reikalavim懦 specifikacija ir projektavimas, naudojant u啪duo膷i懦, komponent懦 ir sek懦 diagramas. Program臈l臈 realizuota naudojant 鈥濳otlin鈥�, 鈥濬irebase鈥�, 鈥濭oogle Maps API鈥� ir i拧testuota pagal testavimo atvejus 鈥� rankiniais ir automatiniais testais. Darb膮 sudaro 拧ios dalys: 寞vadas, kelioni懦 motociklu poreiki懦 analiz臈, egzistuojan膷i懦 kelioni懦 planavimo program臈li懦 analiz臈, technologij懦 analiz臈, reikalavim懦 specifikacija, mobiliosios 辫谤辞驳谤补尘臈濒臈s projektavimas, mobiliosios 辫谤辞驳谤补尘臈濒臈s realizacija, mobiliosios 辫谤辞驳谤补尘臈濒臈s testavimas, i拧vados. Darbo apimtis 鈥� 56 puslapiai be pried懦, 60 paveiksl懦 ir 10 lenteli懦. Rengiant darb膮 buvo panaudota 20 literat奴ros 拧altini懦.

The main objective of the final bachelor thesis is to improve the planning of motorcycle trips by developing a mobile application that automatically generates routes based on selected criteria. In the analytical part, a detailed analysis of trip planning and tracking apps was carried out and the main functionalities and requirements were identified. In the design part, functional and non-functional requirements were identified, and requirements specification and design were carried out using task, component and sequence diagrams. The application was implemented using Kotlin, Firebase, Google Maps API and tested against test cases - manual and automated tests. The thesis consists of the following parts: introduction, motorcycle travel needs analysis, analysis of existing travel planning apps, technology analysis, requirements specification, mobile app design, mobile app implementation, mobile app testing, conclusions. The thesis consists of 56 pages without appendices, 60 figures and 10 tables. 20 literature sources were used in the preparation of the work.

Baigiamajame bakalauro darbe sukurtas realaus laiko vabzd啪i懦 atpa啪inimo ir identifikavimo mobiliosios 辫谤辞驳谤补尘臈濒臈s prototipas. Darbo tikslas 鈥� palengvinti vabzd啪i懦 寞vairov臈s ir pavojingumo vertinim膮, taikant dirbtinio intelekto technologijas. Analitin臈je dalyje ap啪velgtos vabzd啪i懦 identifikavimo problemos, pana拧i懦 program臈li懦 funkcionalumai, objekt懦 atpa啪inimo algoritmai (YOLO, SSD, Faster R-CNN), mobili懦j懦 aplikacij懦 k奴rimo karkasai, serverio infrastrukt奴ros ir dirbtinio intelekto modeliai. Praktin臈je dalyje apra拧yti funkciniai ir nefunkciniai reikalavimai, neuroninio tinklo mokymo procesas, bei sukurto prototipo ekranai. Taip pat pateikti testavimo rezultatai, vertinant 辫谤辞驳谤补尘臈濒臈s funkcionalum膮, bei identifikavimo tikslum膮. Baigiam膮j寞 darb膮 sudaro: 寞vadas, vabzd啪i懦 寞vairov臈s ir identifikavimo i拧拧奴kiai, pana拧i懦 i拧mani懦j懦 program臈li懦 ap啪valga, technologij懦 ap啪valga, esamos ir planuojamos b奴senos analiz臈, esmin臈s u啪duotys, reikalavim懦 specifikavimas, vabzd啪i懦 atpa啪inimo mobiliosios 辫谤辞驳谤补尘臈濒臈s projektin臈 dalis, vabzd啪i懦 identifikavimo modelis, 辫谤辞驳谤补尘臈濒臈s prototipas 鈥濱nsectVision鈥�, E2E testavimas, integraciniai testai, neuroninio tinklo modelio testavimas. Darbo apimtis 鈥� 80 puslapiai be pried懦, 49 iliustracijos, 31 lentel臈 ir 36 literat奴ros 拧altiniai.

This bachelor's thesis contains the development of a real-time insect recognition and identification mobile application prototype. The objective of the thesis is to facilitate the evaluation of insect diversity and potential danger by employing artificial intelligence technologies. The analytical part examines the challenges of insect identification, functionalities of similar mobile applications, object detection algorithms (YOLO, SSD, Faster R-CNN), mobile development frameworks, server infrastructure, and AI models. The practical part presents the functional and non-functional requirements, the neural network training process, and the design of the application prototype. Furthermore, the results of testing are provided, assessing both the application's functionality and identification accuracy. The structure of the thesis includes: an introduction; a discussion of insect diversity and identification challenges; a review of similar smart applications; a technological overview; analysis of the current and proposed system architecture; definition of core tasks; requirements specification; the design of the insect recognition mobile application; the insect identification model; the prototype InsectVision; end-to-end testing; integration testing; and neural network model evaluation. The thesis comprises 80 pages (excluding appendices), 49 illustrations, 31 tables, and 36 references.

艩io darbo tikslas 鈥� atlikti duomen懦 analitikos 寞ranki懦 lyginam膮j膮 analiz臋, siekiant nustatyti j懦 praktin寞 pritaikomum膮 Lietuvos verslo sektoriuje. Darbe analizuojami 拧i懦 寞ranki懦 funkcionalumo aspektai, 寞vertinami j懦 privalumai, tr奴kumai ir pritaikymo galimyb臈s 寞vairiose verslo srityse. Tyrime taikyti 寞vairus analiz臈s metodai ir pasi奴lytas vertinimo modelis, apimantis svarbiausius efektyvumo, funkcionalumo ir patogumo rodiklius. Pagrindiniai tyrimo rezultatai parod臈, kad kiekvienas 寞rankis turi specifini懦 privalum懦, ta膷iau j懦 pasirinkimas priklauso nuo 寞mon臈s poreiki懦 ir veiklos pob奴d啪io. Darbo i拧vados pateikia rekomendacijas, kok寞 寞rank寞 rinktis skirtingoms organizacijoms.

The purpose of this work is to conduct a comparative analysis of data analytics tools to determine their practical applicability in the Lithuanian business sector. The work analyzes the functional aspects of these tools, evaluates their advantages, disadvantages and application possibilities in various business areas. Various analysis methods were applied in the study, and an evaluation model was proposed, encompassing the most important indicators of efficiency, functionality, and usability. The main results of the study showed that each tool has specific advantages, but their choice depends on the needs of the company and the nature of its activities. The conclusions of the work provide recommendations on which tool to choose for different organizations.

艩iame darbe nagrin臈jamas virtuali懦 aplink懦 (VE) naudojimas robot懦 testavimui, ypating膮 d臈mes寞 skiriant sustiprinamojo mokymosi (RL) metodams. Darbe pabr臈啪iama, kad nors VE si奴lo saug懦 ir ekonomi拧k膮 sprendim膮 robot懦 algoritmams bandyti, tikslus realaus pasaulio scenarij懦 atk奴rimas i拧lieka i拧拧奴kiu. Sustiprinamasis mokymasis yra pristatomas kaip perspektyvus b奴das 拧iems i拧拧奴kiams spr臋sti, leid啪iant robotams mokytis i拧 s膮veikos su aplinka. Tyrimo tikslas yra sukurti metodik膮 ir prototip膮, skirt膮 virtualioms robot懦 testavimo aplinkoms, naudojant sustiprinamojo mokymosi algoritmus, siekiant prisid臈ti prie patikimesni懦 ir efektyvesni懦 robot懦 bandymo metodik懦 k奴rimo. Darbo apimtis 58psl. teksto be pried懦, 17 iliustr., 4 lent., 48 bibliografini懦 拧altini懦, 1 priedas.

This thesis explores the use of virtual environments (VE) for robot testing, with a special focus on reinforcement learning (RL) methods. The work emphasizes that although VE offers a safe and cost-effective solution for testing robot algorithms, accurately replicating real-world scenarios remains a challenge. Reinforcement learning is presented as a promising approach to address these challenges, allowing robots to learn from interaction with the environment. The aim of the research is to develop a methodology and a prototype for virtual robot testing environments using reinforcement learning algorithms, to contribute to the development of more reliable and efficient robot testing methodologies. The scope of the work is 58 pages of text without appendices, 17 illustrations, 4 tables, 48 bibliographical sources, 1 appendix.

Baigiamojo bakalauro darbo tikslas buvo sukurti vaist懦 vartojimo priminimo sistem膮 su balso asistentu ir taip padidinti galimybes senjorams bei ne寞galiesiems. Sistema leid啪ia sukurti vaist懦 vartojimo tvarkara拧膷ius, per啪i奴r臈ti pasirinktos dienos priminimus, per啪velgti bei pakeisti praeityje esan膷i懦 priminim懦 b奴sen膮 esant poreikiui, keisti 辫谤辞驳谤补尘臈濒臈s prieinamumo nustatymus, gauti pilno ekrano vaist懦 vartojimo priminimus bei naudotis balso asistentu. Sistema 寞gyvendinta kaip mobilioji 辫谤辞驳谤补尘臈濒臈 Anbdroid operacinei sistemai naudojant React Native mobili懦j懦 program臈li懦 karkas膮. Balso asistentui 寞gyvendinti panaudota react-native-tts teksto 寞 bals膮 biblioteka. Visi duomenys saugomi ir valdomi mobili膮jame 寞renginyje, nenaudojant i拧orin臈s duomen懦 baz臈s norint suteikti galimyb臋 naudotis program臈le be interneto ry拧io.

The aim of this bachelor's thesis was to develop a medication reminder system with a voice assistant, thereby improving accessibility for the elderly and people with disabilities. The system allows users to create medication schedules, view reminders for a selected day, review and modify the status of past reminders if needed, change app accessibility settings, receive full-screen medication notifications, and interact with a voice assistant. The system was implemented as a mobile application for the Android operating system using the React Native mobile app framework. The voice assistant was implemented using the react-native-tts text-to-speech library. All data is stored and managed locally on the mobile device without using an external database, enabling use of the app without an internet connection.

艩iame darbe pasi奴lytas metodas po啪ymi懦 svarbai reguliuoti, siekiant padidinti transporto srauto prognoz臈s tikslum膮 protingose 拧viesofor懦 valdymo sistemose. Tyrimas sprend啪ia ma拧ininio mokymosi modeli懦 nepermatomumo klausim膮, remdamasis ai拧kiomis Explainable Boosting Regressor rezultat懦 interpretacijomis, kurios leid啪ia pagerinti prognozes ir i拧laikyti skaidrum膮. Metodika taiko realius miesto sankry啪懦 transporto detektori懦 duomenis ir pasitelkia i拧sami膮 po啪ymi懦 in啪inerij膮, 寞traukiant laiko, erdv臈s ir transporto sud臈ties kintamuosius. Laiko po啪ymiai naudoja sinuso-kosinuso pakeitimus cikliniams modeliams aptikti, o kategorinio kodavimo metodai analizuoja transporto scenarijus ir transporto priemoni懦 tip懦 paskirstymus. Po啪ymi懦 svarbos koregavimui naudojamas antrasis Ridge regresijos modelis, skirtas prognozi懦 liekanoms, pasitelkiant penkis svarbiausius po啪ymius i拧 interpretacin臈s analiz臈s. Duomen懦 rinkinys sudaro 5,952 寞ra拧懦 i拧 Abdul-haq sankry啪os Kabule, Afganistane, u啪fiksuot懦 15 minu膷i懦 intervalais, atskleid啪iant transporto priemoni懦 skai膷ius keturiose kategorijose. Eksperimentai atskleid啪ia reik拧mingus na拧umo pager臈jimus: 17,08% suma啪臈jus kvadratin臈s paklaidos vidurkiui, 8,94% kritus 拧aknies kvadrato paklaidai ir 13,11% suma啪臈jus absoliu膷iai paklaidai. Po啪ymi懦 analiz臈 parodo, kad transporto sud臈tis lemia 66,5% prognozavimo geb臈jim懦, transporto s膮lygos 鈥� 21,0%, o scenarij懦 po啪ymiai 鈥� 14,0%. Metodas identifikuoja trimodalin寞 dienos transporto model寞 su ryto, popiet臈s ir vakaro pikais.

This thesis introduces a correction framework based on feature importance to boost the accuracy of traffic flow forecasts in intelligent traffic light control systems. The study tackles the lack of transparency in machine learning models by employing the interpretable results from Explainable Boosting Regressor to refine predictions while ensuring clarity. The approach uses real-world data from urban traffic detectors, applying detailed feature engineering with variables related to time, space, and vehicle composition. Cyclical patterns are captured through sine-cosine transformations for temporal features, while categorical encoding manages different traffic scenarios and vehicle type distributions. To correct predictions, a secondary Ridge regression model is trained on the residuals of predictions, utilizing the five key features identified from interpretability analysis. The dataset comprises 5,952 entries from Abdul-haq Crossroad in Kabul, Afghanistan, recording vehicle counts in four categories every 15 minutes. The experiment shows substantial performance enhancements: 17.08% decrease in Mean Squared Error, 8.94% in Root Mean Squared Error, and 13.11% in Mean Absolute Error. Feature analysis indicates that vehicle composition accounts for 66.5% of predictive influence, traffic conditions provide 21.0%, and scenario features contribute 14.0%. The framework also reveals three daily traffic peaks in the morning, afternoon, and evening. This method retains computational efficiency in post-training, rendering it apt for real-time edge deployment in traffic management systems.

Baigiamojo darbo tikslas - pagerinti teisini懦 duokument懦 analiz臋, taikant didel臈s kalbos model寞 teisin臈s informacijos apdorojimui. Analitin臈je dalyje i拧analizuoti didel臈s kalbos modeli懦 informacijos ir teksto apibendinimo procesai. Atvaizduoti esamos ir siekiamos b奴senos vaizdieji paveiksl臈liai. I拧analizuotos esamos teisini懦 dokument懦 apibendrinimo program懦 sistemos. Atlikta technologij懦 skirt懦 sukurti program懦 sistem膮 analiz臈. Projektin臈je dalyje pateikta esmini懦 program懦 sistemos u啪duo膷i懦 diagrama. Apra拧yti funkciniai ir nefunkciniai reikalavimai. Pateiktos sistemos architekt奴ros, komponent懦, klasi懦 ir objekt懦 diagramos. Realizacijos ir testavimo dalyje pateikiamos program懦 sistemos veikimo ekrano nuotraukos ir veikimo apra拧ymas. Pateikiamas sistemos semantin臈s paie拧kos algoritmo pseudo-kodas ir apra拧ymas . Pateikti testavimo scenarijai ir atvejai, kuriais patikrintas funkcini懦 ir nefunkcini懦 reikalavim懦 寞gyvendinimas. Testavimas atliktas rankiniu ir automatiniu b奴du. Pateikiamas sistemos testavimas su nekilnojamo turto teisiniais dokumentais, pateikiamas sistemos ir eksperto analiz臈s palyginimas. Darbo strukt奴ra: 漠vadas, analitin臈 dalis, technologij懦 ap啪valga, sistemos projektavimas, sistemos realizavimas ir testavimas, i拧vados ir si奴lymai, literat奴ros s膮ra拧as ir priedai. Darbo apimtis: 62 puslapiai be pried懦, 28 paveiksl臈liai, 16 lenteli懦 ir 37 拧altiniai.

The objective of this thesis is to improve the analysis of legal documents by applying a large language model for processing legal information. In the analytical part, the processes of information handling and text summarization using large language models are analyzed. Visual representations of the current and desired states are presented. Existing legal document summarization software systems are analyzed. An analysis of technologies suitable for creating the software system is also provided. In the design part, a diagram of the essential tasks of the software system is presented. Functional and non-functional requirements are described. System architecture, component, class, and object diagrams are provided. In the implementation and testing part, screenshots and descriptions of the software system鈥檚 operation are presented. The pseudocode and explanation of the system鈥檚 semantic search algorithm are provided. Testing scenarios and cases are presented to verify the implementation of functional and non-functional requirements. Testing was carried out both manually and automatically. Testing of the system was conducted with legal documents related to real estate. A comparison between the system's and the expert's analysis is provided. Thesis structure: Introduction, analytical part, technology overview, system design, system implementation and testing, conclusions and recommendations, references, and appendices. Thesis consists of 62 pages without appendices, 28 illustrations, 16 tables, and 37 references.