|
Title:
|
CONTEXT-AWARE SHIPBUILDING KNOWLEDGE RECOMMENDATION: EXPLOITING KNOWLEDGE GRAPHS WITH DIFFERENT STRUCTURE |
|
Author(s):
|
Bo Song, Qiao Zhou and Zuhua Jiang |
|
ISBN:
|
978-989-8704-56-6 |
|
Editors:
|
Miguel Baptista Nunes, Pedro Isaías and Philip Powell |
|
Year:
|
2024 |
|
Edition:
|
Single |
|
Keywords:
|
Knowledge Recommendation, Context-Aware, Knowledge Graph, RippleNet |
|
Type:
|
Full |
|
First Page:
|
21 |
|
Last Page:
|
28 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
if you are a member please login 
|
|
Paper Abstract:
|
Knowledge recommendation faces unique challenges in high-tech organizations such as shipbuilding companies. Inactive interactions between users and knowledge items hinder the application of collaborative filtering, while the complexity of knowledge items makes content-based filtering challenging. In this paper, we propose a knowledge recommendation method using knowledge graphs, neural networks, and a context-aware mechanism. Utilizing two types of knowledge graphs: a sparse knowledge graph modelling the categorical relationships between knowledge items and a dense knowledge graph modelling the term-level semantic relatedness between knowledge items, we propagate shipbuilding task context and user-item interactions through the graph to assign interest scores to the knowledge items in order to determine recommendation prioritization. Experiments show that with some surprise, the sparse knowledge graph combined with RippleNet is the best model in terms of knowledge recommendation accuracy. This research points out a new direction of combinatorial optimization of knowledge recommendation and knowledge graph structures, which has agood application prospect in knowledge-intensive industry. |
|
|
|
|
|
|
