Tool for manual or automatic nurikabe solving

V magistrskem delu smo raziskali področje reševanja ugank nurikabe. Implementirali smo algoritem za učinkovito reševanje in prikazali, zakaj je to izjemno težko za človeka kot za računalnik. Dan problem je NP-poln, kar pomeni, da ne obstaja algoritem, ki bi našel rešitev v polinomskem času. Za reševanje ugank smo izdelali namizno orodje v ogrodju .NET, s pomočjo WPF-ja ter programskega jezika C# in spletno aplikacijo s pomočjo tehnologij MongoDB, Node.js in Angular. Uganke lahko rešujemo samostojno, pri čemer lahko zaprosimo program, da nam da nasvet ali pa če obupamo, polje reši namesto nas. Raziskali in implementirali smo tudi reševanja nurikabe ugank s pomočjo nevronskih mrež. Naša orodja in algoritem smo ovrednotili tako kvalitativno kot tudi kvantitativno na naključno izbranih primeri nurikabe polj. Na osnovi rezultatov lahko trdimo, da naš algoritem deluje učinkovito in orodji ponujata primerno uporabniško izkušnjo.In the master’s thesis we researched the field of nurikabe puzzle solving. We implemented an algorithm for solving those efficiently and showing why it is immensely difficult for a human as well as for a computer. The given problem is NP-complete, which means that there is no algorithm that can find a solution in polynomial time. To solve the puzzles, we built a desktop tool in the .NET framework, using WPF and the C# programming language and a web application using MongoDB, Node.js and Angular technologies. The puzzles can be solved either independently, where we can ask the program to give us hints or, if we give up, have it solve the puzzle for us. We also researched and implemented nurikabe solving assistance with neural networks. We evaluated our tools and algorithm both qualitatively and quantitatively on randomly selected samples of nurikabe fields. Based on the results, we can say that our algorithm works efficiently, and the tools offer a proper user experience