A Survey on Personalized PageRank Computation Algorithms

Park, Sungchan; Lee, Wonseok; Choe, Byeongseo; Lee, Sang-goo

doi:10.1109/access.2019.2952653

Cited by 35 publications

(20 citation statements)

References 85 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This algorithm has been applied successfully to a broad range of graph structures in order to find relevant items [28,29] and we include it as a baseline in our analysis. Its performance, however, is hindered by pages with high in-degree that function as hubs and obtain high scores regardless of the starting point.…”

Section: Graph-based Methodsmentioning

confidence: 99%

A general method for estimating the prevalence of Influenza-Like-Symptoms with Wikipedia data

De Toni,

Consonni,

Montresor

2020

Preprint

View full text Add to dashboard Cite

Influenza is an acute respiratory seasonal disease that affects millions of people worldwide and causes thousands of deaths in Europe alone. Being able to estimate in a fast and reliable way the impact of an illness on a given country is essential to plan and organize effective countermeasures, which is now possible by leveraging unconventional data sources like web searches and visits. In this study, we show the feasibility of exploiting information about Wikipedia's page views of a selected group of articles and machine learning models to obtain accurate estimates of influenza-like illnesses incidence in four European countries: Italy, Germany, Belgium, and the Netherlands. We propose a novel language-agnostic method, based on two algorithms, Personalized PageRank and CycleRank, to automatically select the most relevant Wikipedia pages to be monitored without the need for expert supervision. We then show how our model is able to reach state-of-the-art results by comparing it with previous solutions.

show abstract

Section: Graph-based Methodsmentioning

confidence: 99%

A general method for estimating the prevalence of Influenza-Like-Symptoms with Wikipedia data

De Toni,

Consonni,

Montresor

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Next, we first describe how we model these two types of similarity individually and then discuss how we combine them to model the overall similarity function. Graph Structure Similarity Personalized Page Rank (PPR) is a widely adopted tool for measuring the relevance between nodes in graph mining [26,29,30]. More recently, it has also been adopted to improve graph representation learning [24,25].…”

Section: Modeling Node Similaritymentioning

confidence: 99%

Enhancing Graph Contrastive Learning with Node Similarity

Chi¹,

Ma²

2022

Preprint

View full text Add to dashboard Cite

Graph Neural Networks (GNNs) have achieved great success in learning graph representations and thus facilitating various graphrelated tasks. However, most GNN methods adopt a supervised learning setting, which is not always feasible in real-world applications due to the difficulty to obtain labeled data. Hence, graph self-supervised learning has been attracting increasing attention. Graph contrastive learning (GCL) is a representative framework for self-supervised learning. In general, GCL learns node representations by contrasting semantically similar nodes (positive samples) and dissimilar nodes (negative samples) with anchor nodes. Without access to labels, positive samples are typically generated by data augmentation, and negative samples are uniformly sampled from the entire graph, which leads to a sub-optimal objective. Specifically, data augmentation naturally limits the number of positive samples that involve in the process (typically only one positive sample is adopted). On the other hand, the random sampling process would inevitably select false-negative samples (samples sharing the same semantics with the anchor). These issues limit the learning capability of GCL. In this work, we propose an enhanced objective that addresses the aforementioned issues. We first introduce an unachievable ideal objective that contains all positive samples and no false-negative samples. This ideal objective is then transformed into a probabilistic form based on the distributions for sampling positive and negative samples. We then model these distributions with node similarity and derive the enhanced objective. Comprehensive experiments on various datasets demonstrate the effectiveness of the proposed enhanced objective under different settings. INTRODUCTIONGraphs are regarded as a type of essential data structure to represent many real-world data, such as social networks [8,11], transportation networks [37], and chemical molecules [21,43]. Many real-world applications based on these data can be naturally treated as computational tasks on graphs. To facilitate these graph-related tasks, it is essential to learn high-quality vector representations for graphs and their components. Graph neural networks (GNNs) [23,41,44], which generalize deep neural networks to graphs, have demonstrated their great power in graph representation learning,

show abstract

“…In PPR, there is a set of restarting nodes which are the only ones that the random walker can jump to when it gets bored of following links. PPR is still widely used and several works propose ways to optimize its execution [ 58 ].…”

Section: Related Workmentioning

confidence: 99%

Approaches to measure class importance in Knowledge Graphs

et al. 2021

View full text Add to dashboard Cite

The amount, size, complexity, and importance of Knowledge Graphs (KGs) have increased during the last decade. Many different communities have chosen to publish their datasets using Linked Data principles, which favors the integration of this information with many other sources published using the same principles and technologies. Such a scenario requires to develop techniques of Linked Data Summarization. The concept of a class is one of the core elements used to define the ontologies which sustain most of the existing KGs. Moreover, classes are an excellent tool to refer to an abstract idea which groups many individuals (or instances) in the context of a given KG, which is handy to use when producing summaries of its content. Rankings of class importance are a powerful summarization tool that can be used both to obtain a superficial view of the content of a given KG and to prioritize many different actions over the data (data quality checking, visualization, relevance for search engines…). In this paper, we analyze existing techniques to measure class importance and propose a novel approach called ClassRank. We compare the class usage in SPARQL logs of different KGs with the importance ranking produced by the approaches evaluated. Then, we discuss the strengths and weaknesses of the evaluated techniques. Our experimentation suggests that ClassRank outperforms state-of-the-art approaches measuring class importance.

show abstract

A Survey on Personalized PageRank Computation Algorithms

Cited by 35 publications

References 85 publications

A general method for estimating the prevalence of Influenza-Like-Symptoms with Wikipedia data

A general method for estimating the prevalence of Influenza-Like-Symptoms with Wikipedia data

Enhancing Graph Contrastive Learning with Node Similarity

Approaches to measure class importance in Knowledge Graphs

Contact Info

Product

Resources

About