Tampere University of Technology

TUTCRIS Research Portal

Social Approach for Context Analysis: Modelling and Predicting Social Network Evolution using Homophily

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Details

Original languageEnglish
Title of host publicationModeling and Using Context
Subtitle of host publication9th International and Interdisciplinary Conference, CONTEXT 2015, Lanarca, Cyprus, Novemer 2-6,2015. Proceedings
PublisherSpringer International Publishing
Pages513-519
Number of pages7
ISBN (Electronic)978-3-319-25591-0
ISBN (Print)978-3-319-25590-3
DOIs
Publication statusPublished - 2015
Publication typeA4 Article in a conference publication
EventInternational and Interdisciplinary Conference on Modeling and Using Context -
Duration: 1 Jan 1900 → …

Publication series

NameLecture Notes in Artificial Intelligence
Volume9405

Conference

ConferenceInternational and Interdisciplinary Conference on Modeling and Using Context
Period1/01/00 → …

Abstract

Understanding the user’s context is important for mobile applications to provide personalized services. Such context is typically based on the user’s own information. In this paper, we show how social network analysis and the study of the individual in a social network can provide meaningful contextual information. According to the phenomenon of homophily, similar users tend to be connected more frequently than dissimilar. We model homophily in social networks over time. Such models strengthen context inference algorithms, which helps determine future status of the user, resulting in prediction accuracy improvements of up to 118% with respect to a naïve classifier.

Keywords

  • Social network analysis, Homophily, Context Inference

Publication forum classification

Field of science, Statistics Finland