This post contains some pointers to common errors to pay attention to when finalizing your paper’s references.

## Don’t blindly cite arXiv papers

arXiv provides the world with access to the newest scientific developments. Remember to think about the quality of the papers that you reference, in particular, the importance of the peer-review process for science. If you find an article on arXiv you should check if there is a peer-reviewed version published elsewhere. The authoritative version of the paper is not the version on arXiv, rather it is the published peer-reviewed version.

Additionally, the two versions may differ significantly. For example, this is the case with one of the papers that I once discussed in the Text and Multimedia Mining class at Radboud:

Compare for yourself.

## How to find the authoritative version

A feature you can use to find different versions of the same paper is in Google Scholar. Under your search result, there will be a link “All N versions”.

## Don’t blindly trust Google Scholar

Google Scholar (and sister services) are really handy. Every minute spent on handling the bookkeeping for your citations is a minute you cannot spend on research. The service hence saves you valuable time, but at a cost: it makes errors and you need to invest time finding and correcting them.

The paper “Very deep convolutional networks for large-scale image recognition” proposing the VGG16 and VGG19 architectures, is cited more than 27.000 times. When we look it up in Google Scholar, it references arXiv. A simple web search however brings us to the website of the original authors, who recommending citing it:

@InProceedings{Simonyan15,
author       = "Karen Simonyan and Andrew Zisserman",
title        = "Very Deep Convolutional Networks for Large-Scale Image Recognition",
booktitle    = "International Conference on Learning Representations",
year         = "2015",
}


Notes: the notes on ArXiv are taken from a list of papers I keep at https://github.com/sbrugman/deep-learning-papers#a-note-on-arxiv, a highly popular but poorly curated list of papers I keep.