Scholarlink

<p>Customer engagement and relationship management have become critical factors in driving brand loyalty and advocacy in today’s competitive market. This paper presents a novel approach for identifying potential customer advocates using Temporal Knowledge Graph Embeddings (TKGEs). Our model integrates XLNet embeddings, Temporal Convolutional Networks (TCNs), and Conditional Random Fields (CRFs) to dynamically capture and predict customer behavior over time. We construct a temporal knowledge graph from customer interaction data, leveraging the hybrid model to encode temporal dynamics and relational structures effectively. The proposed methodology not only advances the field of customer relationship management (CRM) but also provides practical tools for businesses to foster strong, long-term customer relationships. Our extensive experiments and ablation studies demonstrate significant improvements over traditional models, highlighting the model’s ability to accurately identify and target potential advocates. Additionally, UMAP-based 3D visualizations in TensorBoard’s Projector offer valuable insights into customer engagement patterns, enabling a more understanding of customer behavior. This comprehensive approach paves the way for more sophisticated and effective CRM strategies in an evolving market landscape.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Temporal knowledge graph embeddings for brand advocate prediction in microblogging

  • 期刊论文

  • 发表日期：

    2026年2月19日

查看全文
  • Bilal Abu-Salih,
  • Salihah Alotaibi,
  • Albandari Lafi Alanazi,
  • Basima Elshqeirat,
  • Tomayess Issa,
  • Muder Almiani,
  • Mohammed Aljaafari

摘要

Customer engagement and relationship management have become critical factors in driving brand loyalty and advocacy in today’s competitive market. This paper presents a novel approach for identifying potential customer advocates using Temporal Knowledge Graph Embeddings (TKGEs). Our model integrates XLNet embeddings, Temporal Convolutional Networks (TCNs), and Conditional Random Fields (CRFs) to dynamically capture and predict customer behavior over time. We construct a temporal knowledge graph from customer interaction data, leveraging the hybrid model to encode temporal dynamics and relational structures effectively. The proposed methodology not only advances the field of customer relationship management (CRM) but also provides practical tools for businesses to foster strong, long-term customer relationships. Our extensive experiments and ablation studies demonstrate significant improvements over traditional models, highlighting the model’s ability to accurately identify and target potential advocates. Additionally, UMAP-based 3D visualizations in TensorBoard’s Projector offer valuable insights into customer engagement patterns, enabling a more understanding of customer behavior. This comprehensive approach paves the way for more sophisticated and effective CRM strategies in an evolving market landscape.