{"id":43269,"date":"2023-09-20T07:45:55","date_gmt":"2023-09-20T05:45:55","guid":{"rendered":"https:\/\/www.inovex.de\/?p=43269"},"modified":"2023-09-20T10:31:14","modified_gmt":"2023-09-20T08:31:14","slug":"how-blind-is-a-blind-peer-review-really","status":"publish","type":"post","link":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/","title":{"rendered":"How Blind Is A Blind Peer Review Really?"},"content":{"rendered":"<p>During the process of academic publishing, the double-blind peer review was introduced to conceal the identity of the author of a paper to reduce the bias toward minorities. This should ensure an equal and fair evaluation across all authors, regardless of gender, country of origin, or affiliation.<\/p>\n<p><!--more--><\/p>\n<p>But you might be wondering: isn&#8217;t it possible to recognize and identify the authors based on other factors such as <strong>the content<\/strong>, <strong>writing style<\/strong>, or <strong>references<\/strong>? Based on a self-generated citation network, this article provides various prediction models that could identify patterns and predict the authors&#8216; sensitive information in their selected references.<\/p>\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_82_2 counter-hierarchy ez-toc-counter ez-toc-custom ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\"><p class=\"ez-toc-title\" style=\"cursor:inherit\"><\/p>\n<\/div><nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#What-is-a-Peer-Review\" >What is a Peer Review?<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Effects-of-double-blind-review\" >Effects of double-blind review<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Data-Collection\" >Data Collection<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Prediction-models\" >Prediction models<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Simple-Count-Approach-SCA\" >Simple Count Approach (SCA)<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Random-Forests-RF\" >Random Forests (RF)<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Graph-Neural-Network\" >Graph Neural Network<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#Evaluation\" >Evaluation<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#How-to-solve-this-issue\" >How to solve this issue?<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#References\" >References<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"What-is-a-Peer-Review\"><\/span>What is a Peer Review?<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><span dir=\"ltr\" role=\"presentation\">Academic peer review, also called scholarly peer review, describes <\/span><span dir=\"ltr\" role=\"presentation\">assessing a draft of scientific papers by experts in the same field (peers) to determine <\/span><span dir=\"ltr\" role=\"presentation\">the suitability of publication in a scientific journal or the proceedings of an academic <\/span><span dir=\"ltr\" role=\"presentation\">conference. Single-blind and double-blind peer reviews have been established as the <\/span><span dir=\"ltr\" role=\"presentation\">most common types of academic peer review. The single-blind peer review conceals <\/span><span dir=\"ltr\" role=\"presentation\">the reviewer&#8217;s identity, enabling them to give an honest critique of the <\/span><span dir=\"ltr\" role=\"presentation\">article without the influence exerted by the author. As the author&#8217;s identity is exposed <\/span><span dir=\"ltr\" role=\"presentation\">in a single-blind peer review, the reviewer may become influenced by the author&#8217;s <\/span><span dir=\"ltr\" role=\"presentation\">reputation, gender, or affiliation. This bias could impact the quality of the evaluation <\/span><span dir=\"ltr\" role=\"presentation\">and ultimately lead to inequality in academic publishing.<br \/>\nThe double-blind peer review, widely adopted in the 21th century, eliminates this issue by keeping the identity of both entities, the writer and the reviewer, hidden. Typically, all information associated with the author is removed. This includes the author&#8217;s name, affiliations, and acknowledgments. Further, pronouns that could hint at the author must be removed, such as \u201cin <strong>our<\/strong> previous work [X].\u201c<\/span><\/p>\n<h2><span class=\"ez-toc-section\" id=\"Effects-of-double-blind-review\"><\/span>Effects of double-blind review<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><span dir=\"ltr\" role=\"presentation\">Some case studies and experiments investigated the effect of adopting the double-blind review.<\/span><\/p>\n<ul>\n<li><span dir=\"ltr\" role=\"presentation\">An experiment conducted in 2017 used the 10th Association for Computing Ma<\/span><span dir=\"ltr\" role=\"presentation\">chinery International Conference on Web Search and Data Mining with two different <\/span><span dir=\"ltr\" role=\"presentation\">peer-reviewing groups.<\/span> <span dir=\"ltr\" role=\"presentation\">One group used a single-blind review, whereas the others <\/span><span dir=\"ltr\" role=\"presentation\">applied a double-blind study. The single-blind peers were significantly more likely <\/span><span dir=\"ltr\" role=\"presentation\">to bid for papers by more prestigious authors, universities, and companies than their <\/span><span dir=\"ltr\" role=\"presentation\">counterparts [1].<\/span><\/li>\n<li><span dir=\"ltr\" role=\"presentation\">In a study, when the gender was known to the reviewer, men&#8217;s work was generally <\/span><span dir=\"ltr\" role=\"presentation\">rated higher by men and women. This was not the case when gender was kept hidden<\/span><span dir=\"ltr\" role=\"presentation\">. Also, the journal Behavioral Ecology noticed a significant increase in publications by <\/span><span dir=\"ltr\" role=\"presentation\">female first authors when introducing the double-blind review in 2001 [2,3].<br \/>\n<\/span><\/li>\n<li><span dir=\"ltr\" role=\"presentation\">Previous papers suggest that single-blind peer review is also prone <\/span><span dir=\"ltr\" role=\"presentation\">to nepotism.<\/span> <span dir=\"ltr\" role=\"presentation\">Authors associated with their corresponding peers had a significant <\/span><span dir=\"ltr\" role=\"presentation\">advantage over those without personal ties [3].<br \/>\n<\/span><\/li>\n<\/ul>\n<p><span dir=\"ltr\" role=\"presentation\">In general, double-blind reviews are more critical and attempt to be more objective <\/span><span dir=\"ltr\" role=\"presentation\">by disregarding a person&#8217;s prestige, gender, or relationships, resulting in a <\/span><span dir=\"ltr\" role=\"presentation\">lower acceptance rate than single-blind peer-reviewed papers.<br \/>\n<\/span><\/p>\n<p>However, there exists one key issue with the double-blind peer review. It provides a false sense of security as the reviewer could still identify the authors of a paper with no personal information attached. The American Journal of Public Health survey <span dir=\"ltr\" role=\"presentation\">showed that most reviewers could recognize the author based on their background knowledge and the reference list [4].<\/span><\/p>\n<h2><span class=\"ez-toc-section\" id=\"Data-Collection\"><\/span>Data Collection<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>We extracted the paper IDs from the <a href=\"https:\/\/arxiv.org\/\" target=\"_blank\" rel=\"noopener\">arXive<\/a> repository and additional information from the <a href=\"https:\/\/www.semanticscholar.org\/\" target=\"_blank\" rel=\"noopener\">Semantic Scholar API<\/a>. We limited the data to the computer vision field from 2000 to 2022. Semantic Scholar returns the essential information, including the title, publication year, authors, and the Digital Object Identifier (DOI). The DOI provided us access to the publisher&#8217;s website, on which a web crawler was implemented to extract the affiliation and country for each author. Typically, publishers do not contain information about the author&#8217;s gender. Consequently, <a href=\"https:\/\/genderize.io\/\">genderize.io<\/a>, a Machine Learning model, predicted the author&#8217;s gender for this dataset.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Prediction-models\"><\/span>Prediction models<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3><span class=\"ez-toc-section\" id=\"Simple-Count-Approach-SCA\"><\/span>Simple Count Approach (SCA)<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p><span dir=\"ltr\" role=\"presentation\">The prediction model presented in this chapter is a self-citation approach introduced <\/span><span dir=\"ltr\" role=\"presentation\">by Hill and Provost [5] in 2003. It aims to classify the authors of papers based on the <\/span><span dir=\"ltr\" role=\"presentation\">most frequently occurring features in their references. This approach leverages the idea that authors cite authors with similar characteristics. For example, female authors tend to cite other female authors, as females are more present in their network. Also, if somebody works for Inovex, the odds are high that they could quote somebody else from the same company due to proximity or collaborations. Authors could also cite their work repeatedly to make a point across.<\/span><\/p>\n<h4>SCA: Evaluation<\/h4>\n\n<table id=\"tablepress-57\" class=\"tablepress tablepress-id-57\">\n<thead>\n<tr class=\"row-1\">\n\t<td class=\"column-1\"><\/td><th class=\"column-2\">Gender<\/th><th class=\"column-3\">Country<\/th><th class=\"column-4\">Affiliation<\/th><th class=\"column-5\">Author<\/th>\n<\/tr>\n<\/thead>\n<tbody class=\"row-striping row-hover\">\n<tr class=\"row-2\">\n\t<td class=\"column-1\">Primary<\/td><td class=\"column-2\">0.62<\/td><td class=\"column-3\">0.32<\/td><td class=\"column-4\">0.21<\/td><td class=\"column-5\">0.18<\/td>\n<\/tr>\n<tr class=\"row-3\">\n\t<td class=\"column-1\">Including Co-Authors<\/td><td class=\"column-2\">-<\/td><td class=\"column-3\">0.53<\/td><td class=\"column-4\">0.28<\/td><td class=\"column-5\">0.21<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n<p>The results suggest that the more specific the features, the lower the accuracy. The gender prediction returns an accuracy of 63 % which seems reasonably good. However, a rudimentary model that predicts only males would return an accuracy of 83 % due to a gender imbalance in the dataset. One could predict the primary country 1\/3 of the time and the affiliation and author 1\/5 by simply taking the most occurring feature in the references. If we include the co-authors, the performance increases significantly.<\/p>\n<h5>Impact of Paper Count<\/h5>\n\n<table id=\"tablepress-58\" class=\"tablepress tablepress-id-58\">\n<thead>\n<tr class=\"row-1\">\n\t<td class=\"column-1\"><\/td><th class=\"column-2\">0-19<\/th><th class=\"column-3\">20-39<\/th><th class=\"column-4\">40-59<\/th><th class=\"column-5\">60-79<\/th>\n<\/tr>\n<\/thead>\n<tbody class=\"row-striping row-hover\">\n<tr class=\"row-2\">\n\t<td class=\"column-1\">Country<\/td><td class=\"column-2\">0.28<\/td><td class=\"column-3\">0.33<\/td><td class=\"column-4\">0.35<\/td><td class=\"column-5\">0.36<\/td>\n<\/tr>\n<tr class=\"row-3\">\n\t<td class=\"column-1\">Affiliation<\/td><td class=\"column-2\">0.17<\/td><td class=\"column-3\">0.25<\/td><td class=\"column-4\">0.33<\/td><td class=\"column-5\">0.36<\/td>\n<\/tr>\n<tr class=\"row-4\">\n\t<td class=\"column-1\">Author<\/td><td class=\"column-2\">0.14<\/td><td class=\"column-3\">0.23<\/td><td class=\"column-4\">0.34<\/td><td class=\"column-5\">0.16<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n<p>We divide the authors into buckets containing the number of papers they published previously. It is observable that the higher the paper count of an author, the easier it is for the SCA to predict the desired feature. One outlier exists for bucket 60-79 and the author feature, which could be due to the small number of authors within that bucket.<\/p>\n<p>Overall, this model provides a simple and effective way to determine the author&#8217;s characteristics. However, this model still holds some limitations, as it assumes that (1) an author must have prior publications and (2) have a high tendency to cite these prior publications. Additionally, the Simple Count Approach is biased towards more represented classes which implicitly disadvantages minority classes.<\/p>\n<h3><span class=\"ez-toc-section\" id=\"Random-Forests-RF\"><\/span>Random Forests (RF)<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>Random Forests were chosen due to the adequate predictive power using an ensemble of decision trees. Instead of directly taking the references of the test papers, we use previously published papers by the same authors. That way, the model can identify patterns in the training set and then use the models&#8216; parameters on the test set. As a result, authors with minor self-citation bias could also be identified.<\/p>\n<p><figure id=\"attachment_43316\" aria-describedby=\"caption-attachment-43316\" style=\"width: 694px\" class=\"wp-caption aligncenter\"><a href=\"https:\/\/www.inovex.de\/wp-content\/uploads\/random_forest.svg\"><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-43316\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/random_forest.svg\" alt=\"Random Forest Process\" width=\"694\" height=\"510\" \/><\/a><figcaption id=\"caption-attachment-43316\" class=\"wp-caption-text\">Random Forest Process (illustration based on [6])<\/figcaption><\/figure>When building each decision tree, the Random Forest takes pseudo samples of the training set. Additionally, it performs <em>feature bagging.<\/em> Thus, it takes a random subset of features from the original training set. As a result, features with high predictive power are not present in all trees, reducing the risk of overfitting the training set. For an unseen test set, the Random Forest model then propagates the input to each decision tree, each predicting an individual output. Regarding class prediction, the majority of votes are taken as the model&#8217;s total output. For regression problems, the average could be taken.<\/p>\n<h4>RF: Evaluation<\/h4>\n\n<table id=\"tablepress-59\" class=\"tablepress tablepress-id-59\">\n<thead>\n<tr class=\"row-1\">\n\t<th class=\"column-1\">Country<\/th><th class=\"column-2\">Affiliation<\/th><th class=\"column-3\">Author<\/th>\n<\/tr>\n<\/thead>\n<tbody class=\"row-striping row-hover\">\n<tr class=\"row-2\">\n\t<td class=\"column-1\">0.46<\/td><td class=\"column-2\">0.39<\/td><td class=\"column-3\">0.38<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n<p>The Random Forest model captures relationships between prior publications in the training set and the test data. Compared to the Simple Count Approach, it improves the accuracy by ~20 % in each category. However, we assume that the performance could be further enhanced as Random Forests are not optimized toward graph-structured data:<\/p>\n<ul>\n<li>The features were converted to one-hot encoded features for the RF. This results in a large sparse matrix representation, which could lead to overfitting.<\/li>\n<li>The matrix representation does not consider edge weights or additional node attributes.<\/li>\n<li>The model can not directly capture structural graph information, making it challenging to detect non-linear relationships.<\/li>\n<\/ul>\n<p><span dir=\"ltr\" role=\"presentation\">Consequently, a Graph Neural Network might be a more suitable option as this <\/span><span dir=\"ltr\" role=\"presentation\">model directly employs Deep Learning in the graph representation of the data.<\/span><\/p>\n<h3><span class=\"ez-toc-section\" id=\"Graph-Neural-Network\"><\/span>Graph Neural Network<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p><span dir=\"ltr\" role=\"presentation\">Neural networks have had a lot of success recently. Researchers made significant progress in <\/span><span dir=\"ltr\" role=\"presentation\">object detection or speech recognition with deep learning architectures, such <\/span><span dir=\"ltr\" role=\"presentation\">as convolutional neural networks, recurrent neural networks, or auto-encoders. Yet, <\/span><span dir=\"ltr\" role=\"presentation\">these paradigms do not work in every use case as they are limited to Euclidean data. <\/span><span dir=\"ltr\" role=\"presentation\">By definition, Euclidean data like images or videos can be plotted in an n-dimensional linear<\/span><span dir=\"ltr\" role=\"presentation\"> space. <\/span><\/p>\n<p><span dir=\"ltr\" role=\"presentation\">However, in this case, scientific papers can reference an arbitrary number <\/span><span dir=\"ltr\" role=\"presentation\">of other papers. Therefore, a citation network deals with non-Euclidean data. When dealing with n<\/span><span dir=\"ltr\" role=\"presentation\">on-Euclidean data, convolutional neural networks and recurrent neural networks <\/span><span dir=\"ltr\" role=\"presentation\">reach their limit as they are not designed to handle arbitrary sizes and complex graph <\/span><span dir=\"ltr\" role=\"presentation\">topologies. Additionally, the embeddings in a graph neural network stay consistent across various permutated graphs. Therefore, it utilizes a permutation invariant function that outputs the same vector regardless of the graph&#8217;s permutation and order.\u00a0<\/span><\/p>\n<h4>Graph Convolutional Neural Network (GCN)<\/h4>\n<p>Graph convolutional neural networks are a specification of graph neural networks. GCNs apply convolutions on the graph by learning local feature representations from their neighborhood. This enables the neural networks for each node to learn about the graph&#8217;s structure.<\/p>\n<p><figure id=\"attachment_43353\" aria-describedby=\"caption-attachment-43353\" style=\"width: 587px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-43353\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/gcn_downstream_tasks.svg\" alt=\"gcn_downstream_tasks\" width=\"587\" height=\"549\" \/><figcaption id=\"caption-attachment-43353\" class=\"wp-caption-text\">GCN downstream calculations (Own Illustration based on [7])<\/figcaption><\/figure>The graph convolutional neural network always contains an encoder. The encoder typically consists of multiple layers. One layer corresponds to one message-passing process. In other words, each node aggregates information from the nearby nodes with a single layer. With two layers, it would aggregate the information of nodes two hops away, and eventually, they will learn about the complete graphs&#8216; structure by stacking more layers.<\/p>\n<p>In a single graph convolutional layer, we apply shared permutation invariant functions \\(\\phi (x_{u},X_{N_{u}})\\) over the local neighborhood of the vertex \\(u\\). \\(N_{U}\\) is defined as the neighboring vertices of the node \\(u\\). \\(X_{N_{u}}\\) aggregates the features of the neighborhood of \\(u\\) and is potentially transformed by some function \\(\\psi\\) with a permutation invariant function \\(\\bigoplus\\). The features of node \\(u\\) are then updated by some function \\(\\phi\\). The functions \\(\\phi\\) and \\(\\psi\\) are typically learnable, whereas the nonparametric operation \\(\\bigoplus\\), like mean or max, is not. Note that the nonparametric operation ensures that the result of the message propagation stays the same regardless of the permutation of the graph.<\/p>\n<p>\\[ h_{u} = \\phi (x_{u}, \\bigoplus_{v \\in N_{u}} c_{uv}\\psi (x_{v})) \\]<\/p>\n<p>In the convolutional layer, the features of the neighbors are initialized with a constant weight \\(c_{uv}\\). That specifies the importance of a neighbor to the final aggregation of \\(u\\). Typically, that relates to the structural properties of graph G (e.g., the number of incoming edges of the neighbor v). Finally, the GCN maps each node in the graph to its corresponding node embedding. These embeddings could be used for node\/graph classification or link prediction.<\/p>\n<p>A graph convolutional neural network can also predict the likelihood of edges between two nodes in the graph. Compared with the other downstream task, this prediction requires nonexistent edges of the graph, as it needs to learn the relationship between the scores and negative labels. For this case, the simplest and most common approach of negative sampling is applied: Uniform sampling. This involves randomly selecting edges of the negative graph with equal probability.<\/p>\n<p>Note that link prediction is a binary classification as the negative edges&#8216; labels are 0, and the positive ones are 1. Further, the likelihood of an edge can be computed with their corresponding pair of node embeddings. Common node pair aggregation functions include the cosine similarity or the dot product. The binary cross entropy loss function is typically used to compute the loss between the likelihood and the labels.<\/p>\n<p>After calculating the loss during training, the next step is to update the models&#8216; parameters with backpropagation. This includes calculating the gradient concerning the model&#8217;s parameters. It then determines and updates the parameters according to the negative gradient using optimizers such as the <a href=\"https:\/\/machinelearningmastery.com\/adam-optimization-algorithm-for-deep-learning\/\" target=\"_blank\" rel=\"noopener\">Adam optimizer<\/a>. Once the parameters have been updated, the next batch of training data is processed. This process is repeated until convergence or a specific number of epochs is reached. The resulting model should be capable of predicting missing or future links.<\/p>\n<h4>Heterogeneous Graph Convolutional Neural Network (HGCN)<\/h4>\n<p>Heterogeneous graph convolutional neural networks extend graph convolutional neural networks. The main difference is that HGCNs contain the functionality to operate on heterogeneous graphs. Specifically, it treats each edge and node type differently, applying different convolutional operations. For instance, for the edge tuple (author, writes, paper), the aggregation function uses a convolutional layer that takes the sum of authors and concatenates the papers&#8216; features. In the case of a co-author edge (author, co-author, author), the HGCN could apply a different aggregation function that directly aggregates the features. As a result, it learns different weight matrices for each node and edge type. With the final embeddings of various node and edge types, it decodes the similarity of the node embeddings as before.<\/p>\n<h4>HGCN: Implementation<\/h4>\n<h5>Graph Structure<\/h5>\n<figure id=\"attachment_43369\" aria-describedby=\"caption-attachment-43369\" style=\"width: 726px\" class=\"wp-caption aligncenter\"><a href=\"https:\/\/www.inovex.de\/wp-content\/uploads\/graph_abstraction_model.svg\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-43369\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/graph_abstraction_model.svg\" alt=\"Graph Abstraction\" width=\"726\" height=\"246\" \/><\/a><figcaption id=\"caption-attachment-43369\" class=\"wp-caption-text\">This visualizes the abstraction of the graph.<\/figcaption><\/figure>\n<p>Each feature of the author is represented as a separate entity in the graph. The regular arrows visualize the message propagation layers used during the training process. The dashed lines are the edges that the models aim to predict. Thus, we train four models, each corresponding to one feature.<\/p>\n<h5>HGCN architecture<\/h5>\n<figure id=\"attachment_43373\" aria-describedby=\"caption-attachment-43373\" style=\"width: 606px\" class=\"wp-caption aligncenter\"><a href=\"https:\/\/www.inovex.de\/wp-content\/uploads\/model_architecture.svg\"><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-43373\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/model_architecture.svg\" alt=\"HGCN architecture\" width=\"606\" height=\"313\" \/><\/a><figcaption id=\"caption-attachment-43373\" class=\"wp-caption-text\">HGCN architecture<\/figcaption><\/figure>\n<p><span dir=\"ltr\" role=\"presentation\">Each class of features is converted to a one-hot encoded vector and stored as a feature <\/span><span dir=\"ltr\" role=\"presentation\">vector in each node. During the next step, multiple sum pooling layers aggregate the authors&#8216; features into the authors&#8216; vertices. The model then aggregates the authors&#8216; information with another sum pooling layer and concatenates the result with the one-hot encoded paper IDs. A graph convolutional layer is then applied to the \u201ccites\u201c relation. A slight variation of the standard GCN function was used to avoid information leakage. That function does not regard its feature vector when aggregating its neighboring vectors:\u00a0<\/span><\/p>\n<p>\\[h_i^{(l+1)} = \\sigma (b^{(l)}+ \\Sigma_{j \\in N(i)} \\frac{1}{c_{ji}} h_j^{l}W^{(l)})\\]<\/p>\n<p>Further, a dropout layer ensures the model is not overfitting on the training set. Additionally, the ReLU function adds non-linearity. A final dense layer transforms the final output to the desired output layer to compute the dot product between the papers&#8216; node embedding and the author&#8217;s feature vector.<\/p>\n<h5>HGCN: Evaluation<\/h5>\n<p>&nbsp;<\/p>\n<figure id=\"attachment_43376\" aria-describedby=\"caption-attachment-43376\" style=\"width: 609px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-43376 size-full\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/training_metrics.png\" alt=\"Training Metrics\" width=\"609\" height=\"589\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/training_metrics.png 609w, https:\/\/www.inovex.de\/wp-content\/uploads\/training_metrics-300x290.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/training_metrics-400x387.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/training_metrics-360x348.png 360w\" sizes=\"auto, (max-width: 609px) 100vw, 609px\" \/><figcaption id=\"caption-attachment-43376\" class=\"wp-caption-text\">Train and Validation metrics during training for Author Prediction<\/figcaption><\/figure>\n<p>During training, we utilized a technique called Early Stopping. Generally, this refers to stopping the training process when the validation accuracy diminishes or decreases. In this case, we obtained the model at epoch 330 with a validation accuracy of ~50 % for authorship prediction. That procedure is repeated for every feature. Finally, the HGCN performance looks as follows:<\/p>\n\n<table id=\"tablepress-60\" class=\"tablepress tablepress-id-60\">\n<thead>\n<tr class=\"row-1\">\n\t<th class=\"column-1\">Gender<\/th><th class=\"column-2\">Country<\/th><th class=\"column-3\">Affiliation<\/th><th class=\"column-4\">Author<\/th>\n<\/tr>\n<\/thead>\n<tbody class=\"row-striping row-hover\">\n<tr class=\"row-2\">\n\t<td class=\"column-1\">0.86<\/td><td class=\"column-2\">0.59<\/td><td class=\"column-3\">0.46<\/td><td class=\"column-4\">0.45<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n<h2><span class=\"ez-toc-section\" id=\"Evaluation\"><\/span>Evaluation<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The results suggest that the Blind-Peer Review is, in fact, not blind. A basic counting approach could already identify 1\/5 of all authors assuming that authors would cite themselves the most. We incorporated a statistical Random Forest model, significantly improving the performance to 38 %. A deep learning model, the heterogeneous graph convolutional neural network could identify almost 50 % of all authors. The model&#8217;s predictive power increases as the features get more generic, with ~60 % for country prediction and 86 % for gender prediction.<\/p>\n\n<table id=\"tablepress-61\" class=\"tablepress tablepress-id-61\">\n<thead>\n<tr class=\"row-1\">\n\t<td class=\"column-1\"><\/td><th class=\"column-2\">Gender<\/th><th class=\"column-3\">Country<\/th><th class=\"column-4\">Affiliation<\/th><th class=\"column-5\">Author<\/th>\n<\/tr>\n<\/thead>\n<tbody class=\"row-striping row-hover\">\n<tr class=\"row-2\">\n\t<td class=\"column-1\">SCA<\/td><td class=\"column-2\">0.60<\/td><td class=\"column-3\">0.32<\/td><td class=\"column-4\">0.21<\/td><td class=\"column-5\">0.18<\/td>\n<\/tr>\n<tr class=\"row-3\">\n\t<td class=\"column-1\">RF<\/td><td class=\"column-2\">0.85<\/td><td class=\"column-3\">0.46<\/td><td class=\"column-4\">0.39<\/td><td class=\"column-5\">0.38<\/td>\n<\/tr>\n<tr class=\"row-4\">\n\t<td class=\"column-1\">HGCN<\/td><td class=\"column-2\">0.86<\/td><td class=\"column-3\">0.59<\/td><td class=\"column-4\">0.46<\/td><td class=\"column-5\">0.45<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n<h2><span class=\"ez-toc-section\" id=\"How-to-solve-this-issue\"><\/span>How to solve this issue?<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Some publishers, such as <a href=\"https:\/\/www.journals.elsevier.com\/social-science-and-medicine\/policies\/double-blind-peer-review-guidelines\" target=\"_blank\" rel=\"noopener\"><em>Elsevier<\/em><\/a>, keep the double-blind review but introduced countermeasures to reduce the risk of authorship identification:<\/p>\n<ul>\n<li>Anonymize the author when self-referencing: [Tom 2003] -&gt; [Anonymous 2003]<\/li>\n<li>Remove and limit non-essential self-references<\/li>\n<\/ul>\n<p><span dir=\"ltr\" role=\"presentation\">Though these points mitigate bias during the reviewing process, this could negatively <\/span><span dir=\"ltr\" role=\"presentation\">affect the quality of the review or is challenging to enforce. The publisher must also instruct the reviewers to avoid looking up personal information.<\/span><\/p>\n<p><span dir=\"ltr\" role=\"presentation\">There are other <\/span><span dir=\"ltr\" role=\"presentation\">methodologies in the peer review process that could tackle that issue differently:<\/span><\/p>\n<p><span dir=\"ltr\" role=\"presentation\">For example, in the Open Peer Review, the identity of the author and the writer are <\/span><span dir=\"ltr\" role=\"presentation\">disclosed. The reviewer, however, is obliged to return feedback on the article. This <\/span><span dir=\"ltr\" role=\"presentation\">improves transparency and potential biases as the reviewer needs to reevaluate and <\/span><span dir=\"ltr\" role=\"presentation\">justify their decision to the public [10].<\/span><\/p>\n<p><span dir=\"ltr\" role=\"presentation\">Elsevier employed the Collaborative Peer Review in an experiment in 2014.<\/span> <span dir=\"ltr\" role=\"presentation\">This p<\/span><span dir=\"ltr\" role=\"presentation\">eer review technique allowed authors, editors, and reviewers to openly discuss and interact during the reviewing process<\/span><span dir=\"ltr\" role=\"presentation\">. Ultimately, about 94 % believed the discussion<\/span><span dir=\"ltr\" role=\"presentation\"> led to a better review. On the downside, due to the discussions, this form of review causes the reviewer to put in significantly more effort. Compared to the classical approach, this also takes longer.<\/span><\/p>\n<div class=\"page\" role=\"region\" data-page-number=\"53\" aria-label=\"Page 53\" data-loaded=\"true\" data-listening-for-double-click=\"true\">\n<div class=\"textLayer\">\n<p><span dir=\"ltr\" role=\"presentation\">In conclusion, all peer review methods have advantages and disadvantages. When <\/span><span dir=\"ltr\" role=\"presentation\">selecting a peer review method, editors should carefully evaluate their requirements <\/span><span dir=\"ltr\" role=\"presentation\">and goals. At the same time, it should ensure the quality and integrity of the publication <\/span><span dir=\"ltr\" role=\"presentation\">and provide a fair and transparent evaluation of scientific papers.<\/span><\/p>\n<h2><span class=\"ez-toc-section\" id=\"References\"><\/span>References<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<\/div>\n<\/div>\n<p>[1] <span dir=\"ltr\" role=\"presentation\">A. Tomkins, M. Zhang, and W. D. Heavlin. \u201cReviewer bias in single- versus<\/span><span dir=\"ltr\" role=\"presentation\">double-blind peer review.\u201c In:<\/span> <span dir=\"ltr\" role=\"presentation\">Proceedings of the National Academy of Sciences <\/span><span dir=\"ltr\" role=\"presentation\">114.48 (2017), pp. 12708\u201312713.<\/span> <span dir=\"ltr\" role=\"presentation\">doi<\/span><span dir=\"ltr\" role=\"presentation\">:<\/span> <span dir=\"ltr\" role=\"presentation\">10.1073\/pnas.1707323114<\/span><span dir=\"ltr\" role=\"presentation\">. eprint:<\/span> <span dir=\"ltr\" role=\"presentation\">https:<\/span><span dir=\"ltr\" role=\"presentation\">\/\/www.pnas.org\/doi\/pdf\/10.1073\/pnas.1707323114<\/span><span dir=\"ltr\" role=\"presentation\">.<\/span><\/p>\n<p>[2] <span dir=\"ltr\" role=\"presentation\">R. M. Blank. \u201cThe Effects of Double-Blind versus Single-Blind Reviewing: Ex<\/span><span dir=\"ltr\" role=\"presentation\">perimental Evidence from The American Economic Review.\u201c In:<\/span> <span dir=\"ltr\" role=\"presentation\">The American <\/span><span dir=\"ltr\" role=\"presentation\">Economic Review<\/span> <span dir=\"ltr\" role=\"presentation\">81.5 (1991), pp. 1041\u20131067.<\/span> <span dir=\"ltr\" role=\"presentation\">issn<\/span><span dir=\"ltr\" role=\"presentation\">: 00028282.<\/span><\/p>\n<p>[3] <span dir=\"ltr\" role=\"presentation\">C. Wenner\u00e5s and A. Wold. \u201cNepotism and sexism in peer-review.\u201c In:<\/span> <span dir=\"ltr\" role=\"presentation\">Nature <\/span><span dir=\"ltr\" role=\"presentation\">387.6631 (May 1997), pp. 341\u2013343.<\/span> <span dir=\"ltr\" role=\"presentation\">issn<\/span><span dir=\"ltr\" role=\"presentation\">: 1476-4687.<\/span> <span dir=\"ltr\" role=\"presentation\">doi<\/span><span dir=\"ltr\" role=\"presentation\">: <\/span><span dir=\"ltr\" role=\"presentation\">10.1038\/387341a0<\/span><span dir=\"ltr\" role=\"presentation\">.<\/span><span dir=\"ltr\" role=\"presentation\">52<\/span><\/p>\n<p>[4] <span dir=\"ltr\" role=\"presentation\">S. J. Ceci and D. P. Peters. \u201cHow blind is blind review.\u201c In:<\/span> <span dir=\"ltr\" role=\"presentation\">American Psychologist <\/span><span dir=\"ltr\" role=\"presentation\">39 (1984), pp. 1491\u20131494.<\/span><\/p>\n<p>[5] <span dir=\"ltr\" role=\"presentation\">S. Hill and F. Provost. \u201cThe Myth of the Double-Blind Review? Author Identifica<\/span><span dir=\"ltr\" role=\"presentation\">tion Using Only Citations.\u201c In:<\/span> <span dir=\"ltr\" role=\"presentation\">SIGKDD Explor. Newsl.<\/span> <span dir=\"ltr\" role=\"presentation\">5.2 (Dec. 2003), pp. 179\u2013184. <\/span><span dir=\"ltr\" role=\"presentation\">issn<\/span><span dir=\"ltr\" role=\"presentation\">: 1931-0145.<\/span> <span dir=\"ltr\" role=\"presentation\">doi<\/span><span dir=\"ltr\" role=\"presentation\">:<\/span> <span dir=\"ltr\" role=\"presentation\">10.1145\/980972.981001<\/span><span dir=\"ltr\" role=\"presentation\">.<\/span><\/p>\n<p>[6] <span dir=\"ltr\" role=\"presentation\">Introducing TensorFlow Decision Forests<\/span><span dir=\"ltr\" role=\"presentation\">.<\/span> <span dir=\"ltr\" role=\"presentation\">https:\/\/blog.tensorflow.org\/2021\/05\/<\/span><span dir=\"ltr\" role=\"presentation\">introducing-tensorflow-decision-forests.html<\/span><span dir=\"ltr\" role=\"presentation\">. Accessed: 2023-3-7<\/span><\/p>\n<p>[7] <span dir=\"ltr\" role=\"presentation\">Y. Long, M. Wu, Y. Liu, Y. Fang, C. Kwoh, J. Luo, and X. Li. \u201cPre-training graph <\/span><span dir=\"ltr\" role=\"presentation\">neural networks for link prediction in biomedical networks.\u201c In: <\/span><span dir=\"ltr\" role=\"presentation\">Bioinformatics <\/span><span dir=\"ltr\" role=\"presentation\">(2022).<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>During the process of academic publishing, the double-blind peer review was introduced to conceal the identity of the author of a paper to reduce the bias toward minorities. This should ensure an equal and fair evaluation across all authors, regardless of gender, country of origin, or affiliation.<\/p>\n","protected":false},"author":336,"featured_media":47270,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"ep_exclude_from_search":false,"footnotes":""},"tags":[511,577],"service":[431],"coauthors":[{"id":336,"display_name":"David Le","user_nicename":"dle"},{"id":339,"display_name":"Lea Petters","user_nicename":"lpetters"},{"id":52,"display_name":"Florian Wilhelm","user_nicename":"fwilhelm"}],"class_list":["post-43269","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","tag-artificial-intelligence-2","tag-graph-neural-networks","service-data-science"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.4 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>How Blind Is A Blind Peer Review Really? - inovex GmbH<\/title>\n<meta name=\"description\" content=\"A deep learning approach, specifically a Graph Convolutional Neural Network (GCN) is deployed to identify the authorship of research papers.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/\" \/>\n<meta property=\"og:locale\" content=\"de_DE\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"How Blind Is A Blind Peer Review Really? - inovex GmbH\" \/>\n<meta property=\"og:description\" content=\"A deep learning approach, specifically a Graph Convolutional Neural Network (GCN) is deployed to identify the authorship of research papers.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/\" \/>\n<meta property=\"og:site_name\" content=\"inovex GmbH\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/inovexde\" \/>\n<meta property=\"article:published_time\" content=\"2023-09-20T05:45:55+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-09-20T08:31:14+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1920\" \/>\n\t<meta property=\"og:image:height\" content=\"1080\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"David Le, Lea Petters, Florian Wilhelm\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:image\" content=\"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review-1024x576.png\" \/>\n<meta name=\"twitter:creator\" content=\"@inovexgmbh\" \/>\n<meta name=\"twitter:site\" content=\"@inovexgmbh\" \/>\n<meta name=\"twitter:label1\" content=\"Verfasst von\" \/>\n\t<meta name=\"twitter:data1\" content=\"David Le\" \/>\n\t<meta name=\"twitter:label2\" content=\"Gesch\u00e4tzte Lesezeit\" \/>\n\t<meta name=\"twitter:data2\" content=\"15\u00a0Minuten\" \/>\n\t<meta name=\"twitter:label3\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data3\" content=\"David Le, Lea Petters, Florian Wilhelm\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/\"},\"author\":{\"name\":\"David Le\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/person\\\/ff8a37efd41b2fcdbccebdeeb902f206\"},\"headline\":\"How Blind Is A Blind Peer Review Really?\",\"datePublished\":\"2023-09-20T05:45:55+00:00\",\"dateModified\":\"2023-09-20T08:31:14+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/\"},\"wordCount\":2943,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/double-blind-peer-review.png\",\"keywords\":[\"Artificial Intelligence\",\"Graph Neural Networks\"],\"articleSection\":[\"Analytics\",\"English Content\",\"General\"],\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/\",\"name\":\"How Blind Is A Blind Peer Review Really? - inovex GmbH\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/double-blind-peer-review.png\",\"datePublished\":\"2023-09-20T05:45:55+00:00\",\"dateModified\":\"2023-09-20T08:31:14+00:00\",\"description\":\"A deep learning approach, specifically a Graph Convolutional Neural Network (GCN) is deployed to identify the authorship of research papers.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#breadcrumb\"},\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#primaryimage\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/double-blind-peer-review.png\",\"contentUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/double-blind-peer-review.png\",\"width\":1920,\"height\":1080,\"caption\":\"A grey paper on sea green background with circles marking certain positions in the text. Two sets of eyes oggling the paper.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-blind-is-a-blind-peer-review-really\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"How Blind Is A Blind Peer Review Really?\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#website\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/\",\"name\":\"inovex GmbH\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"de\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#organization\",\"name\":\"inovex GmbH\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/2021\\\/03\\\/inovex-logo-16-9-1.png\",\"contentUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/2021\\\/03\\\/inovex-logo-16-9-1.png\",\"width\":1921,\"height\":1081,\"caption\":\"inovex GmbH\"},\"image\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.facebook.com\\\/inovexde\",\"https:\\\/\\\/x.com\\\/inovexgmbh\",\"https:\\\/\\\/www.instagram.com\\\/inovexlife\\\/\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/inovex\",\"https:\\\/\\\/www.youtube.com\\\/channel\\\/UC7r66GT14hROB_RQsQBAQUQ\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/person\\\/ff8a37efd41b2fcdbccebdeeb902f206\",\"name\":\"David Le\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/2e18a6b9efed86da4be890da2dafb2b5517a63a51ac7ec0683958c79548b1970?s=96&d=retro&r=ga746db7b1c89ba759a177c8d8178638a\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/2e18a6b9efed86da4be890da2dafb2b5517a63a51ac7ec0683958c79548b1970?s=96&d=retro&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/2e18a6b9efed86da4be890da2dafb2b5517a63a51ac7ec0683958c79548b1970?s=96&d=retro&r=g\",\"caption\":\"David Le\"},\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/author\\\/dle\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"How Blind Is A Blind Peer Review Really? - inovex GmbH","description":"A deep learning approach, specifically a Graph Convolutional Neural Network (GCN) is deployed to identify the authorship of research papers.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/","og_locale":"de_DE","og_type":"article","og_title":"How Blind Is A Blind Peer Review Really? - inovex GmbH","og_description":"A deep learning approach, specifically a Graph Convolutional Neural Network (GCN) is deployed to identify the authorship of research papers.","og_url":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/","og_site_name":"inovex GmbH","article_publisher":"https:\/\/www.facebook.com\/inovexde","article_published_time":"2023-09-20T05:45:55+00:00","article_modified_time":"2023-09-20T08:31:14+00:00","og_image":[{"width":1920,"height":1080,"url":"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review.png","type":"image\/png"}],"author":"David Le, Lea Petters, Florian Wilhelm","twitter_card":"summary_large_image","twitter_image":"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review-1024x576.png","twitter_creator":"@inovexgmbh","twitter_site":"@inovexgmbh","twitter_misc":{"Verfasst von":"David Le","Gesch\u00e4tzte Lesezeit":"15\u00a0Minuten","Written by":"David Le, Lea Petters, Florian Wilhelm"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#article","isPartOf":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/"},"author":{"name":"David Le","@id":"https:\/\/www.inovex.de\/de\/#\/schema\/person\/ff8a37efd41b2fcdbccebdeeb902f206"},"headline":"How Blind Is A Blind Peer Review Really?","datePublished":"2023-09-20T05:45:55+00:00","dateModified":"2023-09-20T08:31:14+00:00","mainEntityOfPage":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/"},"wordCount":2943,"commentCount":0,"publisher":{"@id":"https:\/\/www.inovex.de\/de\/#organization"},"image":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#primaryimage"},"thumbnailUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review.png","keywords":["Artificial Intelligence","Graph Neural Networks"],"articleSection":["Analytics","English Content","General"],"inLanguage":"de","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/","url":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/","name":"How Blind Is A Blind Peer Review Really? - inovex GmbH","isPartOf":{"@id":"https:\/\/www.inovex.de\/de\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#primaryimage"},"image":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#primaryimage"},"thumbnailUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review.png","datePublished":"2023-09-20T05:45:55+00:00","dateModified":"2023-09-20T08:31:14+00:00","description":"A deep learning approach, specifically a Graph Convolutional Neural Network (GCN) is deployed to identify the authorship of research papers.","breadcrumb":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#breadcrumb"},"inLanguage":"de","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/"]}]},{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#primaryimage","url":"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review.png","contentUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/double-blind-peer-review.png","width":1920,"height":1080,"caption":"A grey paper on sea green background with circles marking certain positions in the text. Two sets of eyes oggling the paper."},{"@type":"BreadcrumbList","@id":"https:\/\/www.inovex.de\/de\/blog\/how-blind-is-a-blind-peer-review-really\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.inovex.de\/de\/"},{"@type":"ListItem","position":2,"name":"How Blind Is A Blind Peer Review Really?"}]},{"@type":"WebSite","@id":"https:\/\/www.inovex.de\/de\/#website","url":"https:\/\/www.inovex.de\/de\/","name":"inovex GmbH","description":"","publisher":{"@id":"https:\/\/www.inovex.de\/de\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.inovex.de\/de\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"de"},{"@type":"Organization","@id":"https:\/\/www.inovex.de\/de\/#organization","name":"inovex GmbH","url":"https:\/\/www.inovex.de\/de\/","logo":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/www.inovex.de\/de\/#\/schema\/logo\/image\/","url":"https:\/\/www.inovex.de\/wp-content\/uploads\/2021\/03\/inovex-logo-16-9-1.png","contentUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/2021\/03\/inovex-logo-16-9-1.png","width":1921,"height":1081,"caption":"inovex GmbH"},"image":{"@id":"https:\/\/www.inovex.de\/de\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/inovexde","https:\/\/x.com\/inovexgmbh","https:\/\/www.instagram.com\/inovexlife\/","https:\/\/www.linkedin.com\/company\/inovex","https:\/\/www.youtube.com\/channel\/UC7r66GT14hROB_RQsQBAQUQ"]},{"@type":"Person","@id":"https:\/\/www.inovex.de\/de\/#\/schema\/person\/ff8a37efd41b2fcdbccebdeeb902f206","name":"David Le","image":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/secure.gravatar.com\/avatar\/2e18a6b9efed86da4be890da2dafb2b5517a63a51ac7ec0683958c79548b1970?s=96&d=retro&r=ga746db7b1c89ba759a177c8d8178638a","url":"https:\/\/secure.gravatar.com\/avatar\/2e18a6b9efed86da4be890da2dafb2b5517a63a51ac7ec0683958c79548b1970?s=96&d=retro&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/2e18a6b9efed86da4be890da2dafb2b5517a63a51ac7ec0683958c79548b1970?s=96&d=retro&r=g","caption":"David Le"},"url":"https:\/\/www.inovex.de\/de\/blog\/author\/dle\/"}]}},"_links":{"self":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts\/43269","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/users\/336"}],"replies":[{"embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/comments?post=43269"}],"version-history":[{"count":6,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts\/43269\/revisions"}],"predecessor-version":[{"id":47272,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts\/43269\/revisions\/47272"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/media\/47270"}],"wp:attachment":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/media?parent=43269"}],"wp:term":[{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/tags?post=43269"},{"taxonomy":"service","embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/service?post=43269"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/coauthors?post=43269"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}