Efficient Behavior Prediction Based on User Events
In 2020 we have witnessed the dawn of machine learning enabled user experience. Now we can predict how users will use an application. Research progressed beyond recommendations, and we are ready to predict user events. Whenever a human interacts with a system, user events are dispatched. They can be...
| Permalink: | http://skupni.nsk.hr/Record/nsk.NSK01001163168/Details |
|---|---|
| Matična publikacija: |
Journal of communications software and systems (Online) 17 (2021), 2 ; str. 134-142 |
| Glavni autori: | Szabo, Peter (Author), Genge, Bela |
| Vrsta građe: | e-članak |
| Jezik: | eng |
| Online pristup: |
https://doi.org/10.24138/jcomss-2020-0011 Elektronička verzija članka Elektronička verzija članka |
| LEADER | 02231naa a22003374i 4500 | ||
|---|---|---|---|
| 001 | NSK01001163168 | ||
| 003 | HR-ZaNSK | ||
| 005 | 20230213112915.0 | ||
| 006 | m d | ||
| 007 | cr|||||||||||| | ||
| 008 | 230213s2021 ci |o |0|| ||eng | ||
| 024 | 7 | |2 doi |a 10.24138/jcomss-2020-0011 | |
| 035 | |a (HR-ZaNSK)001163168 | ||
| 040 | |a HR-ZaNSK |b hrv |c HR-ZaNSK |e ppiak | ||
| 041 | 0 | |a eng | |
| 042 | |a croatica | ||
| 044 | |a ci |c hr | ||
| 080 | 1 | |2 2011 | |
| 100 | 1 | |a Szabo, Peter |4 aut |9 HR-ZaNSK | |
| 245 | 1 | 0 | |a Efficient Behavior Prediction Based on User Events |h [Elektronička građa] |c Peter Szabo, Bela Genge. |
| 300 | |b Ilustr. | ||
| 504 | |a Bibliografija: | ||
| 504 | |a Summary. | ||
| 520 | |a In 2020 we have witnessed the dawn of machine learning enabled user experience. Now we can predict how users will use an application. Research progressed beyond recommendations, and we are ready to predict user events. Whenever a human interacts with a system, user events are dispatched. They can be as simple as a mouse click on a menu item or more complex, such as buying a product from an eCommerce site. Collaborative filtering (CF) has proven to be an excellent approach to predict events. Because each user can generate many events, this inevitably leads to a vast number of events in a dataset. Unfortunately, the operation time of CF increases exponentially with the increase of data-points. This paper presents a generalized approach to reduce the dataset"s size without compromising prediction accuracy. Our solution transformed a dataset containing over 20 million user events (20,692,840 rows) into a sparse matrix in about 7 minutes (434.08 s). We have used this matrix to train a neural network to accurately predict user events. | ||
| 700 | 1 | |a Genge, Bela |4 aut |9 HR-ZaNSK | |
| 773 | 0 | |t Journal of communications software and systems (Online) |x 1846-6079 |g 17 (2021), 2 ; str. 134-142 |w nsk.(HR-ZaNSK)000644741 | |
| 981 | |b Be2021 | ||
| 856 | 4 | 0 | |u https://doi.org/10.24138/jcomss-2020-0011 |
| 856 | 4 | 0 | |u https://jcoms.fesb.unist.hr/10.24138/jcomss-2020-0011/ |y Elektronička verzija članka |
| 856 | 4 | 0 | |u https://jcoms.fesb.unist.hr/pdfs/v17n2_2020-0011_szabo.pdf |y Elektronička verzija članka |
| 856 | 4 | 1 | |y Digitalna.nsk.hr |