ProdInfluencerNet: A Novel Product-Centric Influencer Recommendation Framework Based on Heterogeneous Networks
Keywords: Influencer Marketing, Influecner Recommnedation, Heterogeneous Information Network, Inductive Learning, ProdInfluencerNet
Abstract: With the proliferation of social media, influencer marketing has emerged as a popular strategy for brands to promote their products. Recent studies have increasingly explored the use of machine learning to recommend suitable influencers for brands. This typically involves analyzing the compatibility of influencer profiles with brand attributes. However, for brands entering new markets or promoting products in unfamiliar categories, existing solutions may be limited due to insufficient information for accurate compatibility matching.
In this paper, we propose ProdInfluencerNet (PIN), a product-centric framework designed for influencer recommendation. PIN effectively models the complex relationships between brands, products, and influencers using Heterogeneous Information Networks (HINs). We categorize sponsored post images using the Google Taxonomy through image classification techniques. By leveraging the taxonomy's hierarchical structure and adopting an inductive learning approach, PIN can accurately recommend influencers for brands, even in new markets or with innovative products. We validate PIN's effectiveness and superiority over existing methods using two Instagram datasets. Furthermore, our analysis reveals that text features in profiles are more critical than images for identifying cooperative relationships between product categories and influencers.
Primary Area: learning on graphs and other geometries & topologies
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 5725
Loading