PredictionMiner: mining the latest individual behavioral rules for personalized contextual pattern predictions

SOFT COMPUTING(2023)

引用 0|浏览0
暂无评分
摘要
The user’s behavior towards smartphone is not static and changes dynamically as the real world changes, depending on various contextual patterns. Finding the latest behavioral rules is challenging because the smartphone log is incremented with the user’s behavior, which is not static and changes daily. Previously, for mining, the latest user behavioral rules researchers have used the recent log of static periods and considered it as the latest behavioral log; however those approaches create accuracy and reliability problems because with time behavioral log keeps on updating and some user behaviors become outdated. On the basis of user’s volatile behaviors toward smartphones, this study devises the issue of modeling an individual’s up-to-date behavioral rules with their smart-phone interaction co-occurring patterns by incorporating the dynamically changing log data. Proposed behavioral-based approach named “PredictionMiner” firstly, mines the dynamic log period which holds the latest behavior of individual users by neglecting the outdated behaviors. Secondly, it extracts the individual latest smartphone machine learning rules with co-occurring contextual patterns. By utilizing individualized co-occurring patterns with the corresponding behavioral rules, the personalized context aware prediction model is built for predicting future smartphone contextual behavioral activities. The proposed approach dynamically mines the latest machine learning rules and removes the outdated rules making it more effective. To make this approach more relevant, real-world contextual notifications dataset has been used. Our experiments and comparisons on each contextual dataset show that proposed rule discovery approach is more adequate and accurate than base models.
更多
查看译文
关键词
Machine learning,Context-aware prediction,Behavioral modeling,Co-occurring patterns mining,Association rule mining
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要