An Approach of Analyzing Movement Patterns Using Word Embeddings from Geo-tagged Tweets

DOI
  • Hirota Masaharu
    Department of Information Science, Okayama University of Science
  • Oda Tetsuya
    Department of Information and Computer Engineering, Okayama University of Science

Description

Many people share content about daily events on social media sites. Understanding people's movement patterns using the contents benefits numerous applications, such as tourism recommendation, city planning, and geo-targeting. This study proposes a new approach for clustering user trajectories to discover movement patterns. Our approach generates feature vectors for movement patterns using the word embedding model to learn movements between a pair of areas quantized by their latitude and longitude. Then, our approach uses multiple models to learn each area of different sizes and integrates the generated embedding vectors. The vectors represent the relationships between the movements from one area to the next. We demonstrated that clustering results by our proposed method.

Journal

Keywords

Details 詳細情報について

  • CRID
    1390854717504302080
  • DOI
    10.5954/icarob.2022.os1-2
  • ISSN
    21887829
  • Text Lang
    en
  • Data Source
    • JaLC
    • Crossref
    • OpenAIRE
  • Abstract License Flag
    Disallowed

Report a problem

Back to top