Neural PathSim for Inductive Similarity Search in Heterogeneous Information Networks

Xiao, Wenyi; Zhao, Huan; Zheng, Vincent W.; Song, Yangqiu

doi:10.1145/3459637.3482454

Cited by 5 publications

(1 citation statement)

References 38 publications

(26 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These characteristics align well with the requirements for CDA prediction. And in capturing local depth information, subgraph structure can obtain more detailed information than node and path 47,48 . Inspired by the aforementioned article 49 , this paper proposes a recursive approach to constructing a GNN subgraph model.…”

Section: B Recursive Creation Of Graph Neural Networkmentioning

confidence: 99%

KGRACDA: A Model Based on Knowledge Graph from Recursion and Attention Aggregation for CircRNA-disease Association Prediction

Wang,

Ma,

Xie

et al. 2023

Preprint

View full text Add to dashboard Cite

SUMMARYCircRNA has been found to exhibit with human diseases, rendering the investigation of the circRNA-disease association (CDA) of significant importance. Deep learning provides a novel avenue for predicting CDA, however, due to the limited availability of comprehensive CDA datasets, most existing methods operate by directly reasoning over triplets of embedded entities and relations, neglecting the explicit capture local depth information. In this study, we propose a novel approach called Knowledge Graph-Enhanced Recursive Aggregation for CDA (KGRACDA), which enables the querying of explicit local structures surrounding triplets, thereby serving as crucial evidence for predicting CDA. First, we construct a comprehensive CDA knowledge graph utilizing a large-scale, multi-source heterogeneous dataset. Subsequently, we develop a graph neural network using recursive techniques and employ attention aggregation to generate CDA prediction scores. Experimental results demonstrate that KGRACDA surpasses other state-of-the-art methods by explicitly capturing local depth information related to CDA. To improve the convenience of CDA prediction, we have updated an interactive web platform named HNRBase v2.0, which provides rich visualization of circRNA data, and allows users to easily download data and predict CDA using KGRACDA. The code and datasets are publicly available athttps://github.com/maoyuanma/KGRACDA.

show abstract

Section: B Recursive Creation Of Graph Neural Networkmentioning

confidence: 99%

KGRACDA: A Model Based on Knowledge Graph from Recursion and Attention Aggregation for CircRNA-disease Association Prediction

Wang,

Ma,

Xie

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

Semantic enhanced Top-k similarity search on weighted HIN

Zhang

et al. 2022

Neural Comput & Applic

View full text Add to dashboard Cite

Similarity searches on heterogeneous information networks (HINs) have attracted wide attention from both industrial and academic areas in recent years; for example, they have been used for friend detection in social networks and collaborator recommendation in coauthor networks. The structural information on the HIN can be captured by multiple metapaths, and people usually utilize metapaths to design methods for similarity search. The rich semantics in HINs are not only structural information but also content stored in nodes. However, the content similarity of nodes was usually not valued in the existing methods. Although some researchers have recently considered both types of information in machine learning-based methods for similarity search, they have used structure and content information separately. To address this issue by balancing the influence of structure and content information flexibly in the process of searching, we propose a double channel convolutional neural network model for top-k similarity search, which uses path instances as model inputs and generates structure and content embeddings for nodes based on different metapaths. We design an attention mechanism to enhance the differences in metapaths for each node. Another attention mechanism is used to combine the content and structure information of nodes. Finally, an importance evaluation function is designed to improve the accuracy and make the model more explainable. The experimental results show that our search algorithm can effectively support top-k similarity search in HINs and achieve higher performance than existing approaches.

show abstract