CHSPAM: a multi-domain model for sequential pattern discovery and monitoring in contexts histories

Context-aware applications adapt their functionalities based on users contexts. Complementarily, a context history has information about previous contexts visited by a user. Context history enables applications to explore users past behavior. Researchers have studied different ways to analyze these...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Pattern analysis and applications : PAA Ročník 23; číslo 2; s. 725 - 734
Hlavní autoři: Dupont, Daniel, Barbosa, Jorge Luis Victória, Alves, Bruno Mota
Médium: Journal Article
Jazyk:angličtina
Vydáno: London Springer London 01.05.2020
Springer Nature B.V
Témata:
Computer Science
Context
Data analysis
Domains
Evolution
Learning
Monitoring
Object recognition
Pattern analysis
Pattern Recognition
Theoretical Advances
Ubiquitous computing
Context histories
Pattern discovery
Data mining
ISSN:1433-7541, 1433-755X
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Context-aware applications adapt their functionalities based on users contexts. Complementarily, a context history has information about previous contexts visited by a user. Context history enables applications to explore users past behavior. Researchers have studied different ways to analyze these data. This article addresses a specific type of data analysis in contexts histories, which is the discovery and monitoring of sequential patterns. The article proposes a model, called CHSPAM, that allows the discovery of sequential patterns in contexts histories databases and keeps track of these patterns to monitor their evolution over time. There are two main contributions of this work. The first one is the use of a generic representation for stored context information on pattern recognition field, which enables the model to be used for different research domains. The second contribution is the fact that CHSPAM monitors discovered pattern evolution over time. We have build a functional prototype that allowed us to conduct experiments in two different applications. The first experiment used the model to perform pattern analysis and evaluate the prediction based on monitored sequential patterns. Prediction accuracy increased by up to 17% when compared to the use of common sequential patterns. On the second experiment, CHSPAM was used as a component of a learning object recommendation application. The application was able to recommend learning objects related to students interests based on monitored sequential patterns extracted from users session history. Usefulness for recommendations reached 84%.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1433-7541
1433-755X
DOI:10.1007/s10044-019-00829-9