Interest Portrait Construction and Dynamic Evolution Analysis Based on Multi-Source Behavioral Data
Journal of Intelligence and Knowledge Engineering · 2026-06 · 4 (2) · 117-124
DOI: 10.62517/jike.202604217 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Intelligence and Knowledge Engineering, 2026-06. Verfasst von Shuaizhe Huang. Die Arbeit nennt 14 Literaturangaben.
Abstract (Originalsprache)
In the digital era, diverse and heterogeneous user behavior data are trapped in data silos. Traditional static or single-source interest profiling methods cannot support precise user modeling. This paper focuses on constructing user interest portraits and analyzing their dynamic evolution using multi-source heterogeneous behavioral data. It integrates e-commerce behavioral logs (from the Taobao platform) and movie rating interaction data (from the MovieLens dataset), constructs a unified behavior representation via data cleaning, alignment, and fusion, uses NLP to extract fine-grained interest tags, and introduces a time-series framework with a sliding window and attention mechanism to capture dynamic features and identify persistence, mutation, and periodic patterns of user interests.
Zitieren
APA
Huang, S. (2026). Interest Portrait Construction and Dynamic Evolution Analysis Based on Multi-Source Behavioral Data. Journal of Intelligence and Knowledge Engineering, 4(2), 117-124. https://doi.org/10.62517/jike.202604217
Vancouver
Huang S. Interest Portrait Construction and Dynamic Evolution Analysis Based on Multi-Source Behavioral Data. Journal of Intelligence and Knowledge Engineering. 2026;4(2):117-124. doi:10.62517/jike.202604217
BibTeX
@article{Huang2026,
author = {Huang, Shuaizhe},
title = {Interest Portrait Construction and Dynamic Evolution Analysis Based on Multi-Source Behavioral Data},
journal = {Journal of Intelligence and Knowledge Engineering},
year = {2026},
volume = {4},
number = {2},
pages = {117--124},
doi = {10.62517/jike.202604217}
}
EUVIMED