LLM-Augmented Knowledge Graph Reasoning with Disentangled Multimodal Fusion for Explainable E-Commerce Conversion Predictio

Abstract excerpt

<title>Abstract</title> <p> E-commerce conversion prediction is a foundation of modern recommender systems, yet existing methods often suffer from three intertwined limitations: entangled multimodal representations that mix spurious and causal factors, shallow knowledge-graph reasoning whose attention weights are not faithfully explanatory, and a lack of human-readable rationales suitable for business decision m...

Topics

Open a Topic to create a Post that cites this publication.

• Algorithms
• Base Sequence
• Chromosome Mapping
• DNA Mutational Analysis
• Genomic Islands
• Haplotypes
• Molecular Sequence Data
• Sequence Analysis, DNA

Identifiers and source

Literature Corpus work

daeb2998-263b-55eb-be0f-a5eedb28a4f9

DOI

10.21203/rs.3.rs-9923118/v1