User action prediction with neural networks
Tunturi, Eetu (2023)
2023
Tieto- ja sähkötekniikan kandidaattiohjelma - Bachelor's Programme in Computing and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2023-05-23
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202305225988
https://urn.fi/URN:NBN:fi:tuni-202305225988
Tiivistelmä
Artificial intelligence and machine learning have been used to solve a wide variety of complex problems in different fields. In this study, machine learning will be used to predict user actions in software applications. The ability to predict user actions would enable the creation of dynamic user interfaces. Dynamic user interfaces can be used to improve the usability of software applications. Usability means that the user can achieve their goals more effectively and efficiently. Additionally, good usability helps new users to learn to use the application faster.
This thesis aimed to study if neural networks can learn patterns in software application usage. Software application usage can be presented as a sequence of individual actions performed by the user. Neural networks were used to predict the user's actions based on their previous actions. Additionally, the performance of different neural network architectures was compared. First, a small review of previous research into user action prediction is provided. In the experimental part of this thesis, neural networks were trained to predict the next element in the sequence of actions. The neural networks were trained using data from Vertex G4, a computer-aided design (CAD) application. The experiments use data from a desktop application with a graphical user interface, but the results also generalize to other kinds of user interfaces, like mobile applications and command-line applications. Three different neural network architectures were experimented with: long short-term memory (LSTM), gated recurrent unit (GRU), and transformer.
The results of this study show that neural networks can be used to predict user actions in certain kinds of software applications. All of the studied architectures achieve similar performance metrics, which shows that they are all valid solutions for predicting user actions. However, the models trained in this study leave a lot of room for improvement. As such, to conclude this study, some ideas for improving the models are presented.
This thesis aimed to study if neural networks can learn patterns in software application usage. Software application usage can be presented as a sequence of individual actions performed by the user. Neural networks were used to predict the user's actions based on their previous actions. Additionally, the performance of different neural network architectures was compared. First, a small review of previous research into user action prediction is provided. In the experimental part of this thesis, neural networks were trained to predict the next element in the sequence of actions. The neural networks were trained using data from Vertex G4, a computer-aided design (CAD) application. The experiments use data from a desktop application with a graphical user interface, but the results also generalize to other kinds of user interfaces, like mobile applications and command-line applications. Three different neural network architectures were experimented with: long short-term memory (LSTM), gated recurrent unit (GRU), and transformer.
The results of this study show that neural networks can be used to predict user actions in certain kinds of software applications. All of the studied architectures achieve similar performance metrics, which shows that they are all valid solutions for predicting user actions. However, the models trained in this study leave a lot of room for improvement. As such, to conclude this study, some ideas for improving the models are presented.
Kokoelmat
- Kandidaatintutkielmat [8935]