A Simple Relational Classifier

By Sofus A. Macskassy, Foster J. Provost. 2003

302 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Evaluation & Benchmarks

Citations in the corpus

DeepWalk: online learning of social representations (2014, methodology): “Relational classification (or collective classification) methods [15, 25, 32] use links between data items as part of the”
Deep Neural Networks for Learning Graph Representations (2016, methodology): “The learned vertex representations were shown to be effective across a few tasks, outperforming several previous approaches such as spectral clustering and modularity method (Tang and Liu 2009a; 2009b; 2011; Macskassy and Provost 2003).”
Graph based anomaly detection and description: a survey (2014, background): “[Macskassy and Provost, 2003] propose a simple baseline algorithm called (probabilistic) weightedvote relational network (wv-RN) classifier where they use only the class labels of objects for classification; they infer the class label of an object by taking a weighted average of the potentially…”
Leveraging Label-Independent Features for Classification in Sparsely Labeled Networks: An Empirical Study (2008, methodology): “wvRN is the weighted-vote relational neighbor classifier [19,7].”
Classification in Networked Data: A Toolkit and a Univariate Case Study (2007, unknown): “One particular very simple and very effective technique highlights the close correspondence between several types of methods introduced for different pur-1 poses: “node-centric” methods, which focus on each node individually; random-ﬁeld methods; and classic connectionist methods.”
State-of-the-Art Deep Learning: Evolving Machine Intelligence Toward Tomorrow’s Intelligent Network Traffic Control Systems (2017, background): “Relational Neighbor (wvRN) [204] for social networks like BlogCatalog, Flickr, and YouTube.”
Structural Neighborhood Based Classification of Nodes in a Network (2016, background): “Macskassy [13] showed that instead of counting votes, taking a weighted vote of each neighbor is more useful.”
Link-Based Classification (2003, unknown): “Next we explore the more general setting 4 This model was inspired by results in [ 21 ].”
Scalable Learning of Collective Behavior (2012, background): “It is shown that [9] a simple weighted vote relational neighborhood classifier [25] works reasonably well on some benchmark relational data and is recommended as a baseline for comparison.”
Leveraging social media networks for classification (2011, methodology): “It is shown in [29] that a simple weighted-vote relational neighborhood classifier [ 28 ] based on collective inference yields almost identical performance as the Gaussian field for semi-supervised learning [55].”
Corporate residence fraud detection (2014, methodology): “Equation (1) presents the weighted-voted Relational Neighbor (wvRN) inference method [21]; with wvRN, the class probability of a node in the graph (a foreign company) is equal to the weighted average probabilities of all of its neighbors (j ∈ N(xi)).”
Local Modeling of Attributed Graphs: Algorithms and Applications (2016, methodology): “1 Relational Learning Relational classification (or collective classification) methods [52, 83, 102] use links between data items as part of the classification process.”
Graph Regularized Transductive Classification on Heterogeneous Information Networks (2010, background): “– Learning with Local and Global Consistency (LLGC) [21] – Weighted-vote Relational Neighbor classifier (wvRN) [10] [11] – Network-only Link-based classification (nLB) [9] [11]”
Opinion Fraud Detection in Online Reviews by Network Effects (2013, methodology): “We modify two algorithms for network classification to handle signed networks for our fraud detection setting: weighted-vote relational classifier (wv-RC) (Macskassy and Provost 2003; Hill, Provost, and Volinsky 2007) and hubsand-authorities (HITS) algorithm (Kleinberg 1998).”
It's who you know: graph mining using recursive structural features (2011, methodology): “In addition, as a baseline, we include a standard homophily-based relational neighbor classi.er: wvRN (short for weighted-vote Relational Neighbor) [23].”
Who cares about your Facebook friends? Credit scoring for microfinance (2015, methodology): “Based on the empirical results from the study, we focus our attention to the following techniques: (i) the tangens hyperbolicum function for creating the top node weights, where we down-weight the very popular items as providing less information for the target variable, (ii) sum of the shared nodes ”