https://github.com/oowekyala/citegraph
A citation network explorer
https://github.com/oowekyala/citegraph
citation-network graphviz
Last synced: 17 days ago
JSON representation
A citation network explorer
- Host: GitHub
- URL: https://github.com/oowekyala/citegraph
- Owner: oowekyala
- License: mit
- Created: 2020-06-30T05:10:47.000Z (almost 5 years ago)
- Default Branch: master
- Last Pushed: 2023-05-26T11:18:02.000Z (almost 2 years ago)
- Last Synced: 2025-04-10T04:06:00.656Z (17 days ago)
- Topics: citation-network, graphviz
- Language: Python
- Homepage:
- Size: 319 KB
- Stars: 21
- Watchers: 3
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Citegraph
Generates easily readable citation graphs. This uses the contents
of a bibtex file to determine what you're interested in.
This uses the semanticscholar API to fetch references for articles.
Since the API is rate-limited and requests are very slow, request
results are cached locally, and the exploration algorithm is engineered
to make every request count. This is done by computing a degree of interest (DOI)
for each known paper, and fetching only the papers we think will
improve the graph the most. See [here](https://github.com/oowekyala/citegraph/blob/master/src/citegraph/explore.py#L10)
for an explanation of the DOI calculation.
### Installation
* Make sure you have Python 3.6+
* Also make sure you have [Graphviz](https://www.graphviz.org/download/) on your path
* Download or clone this repo
```shell script
$ git clone https://github.com/oowekyala/citegraph.git && cd citegraph
```
* Make sure you have all the required Python packages:
```shell script
$ python3 -m pip install -r DEPENDENCIES
```
* Add the `bin` directory to you PATH, or just use `bin/citegraph`
### Usage
Find out the ID of an interesting paper on [semanticscholar.org](https://www.semanticscholar.org/),
see for example the highlighted section of this picture:
Remove spaces, and you can pass that ID directly to `citegraph`:
```shell script
$ bin/citegraph CorpusID:125964925
[1 / 80 / 145] (DOI 0.0) Fractal calculus and its geometrical explanation
[2 / 80 / 145] (DOI 1.25) Fractal approach to heat transfer in silkworm cocoon hierarchy
...
[80 / 80 / 5602] (DOI 1.428) Bubble Electrospinning for Mass Production of Nanofibers
Hit max size threshold
Rendering...
Rendered to graph.pdf
```
Here's what the graph would look like (using a max `--size` of 20 for readability):
Use `citegraph --help` to find out about all the options.
### Exploration parameters
By default the exploration algorithm is biased towards exploring
downward links. To also explore the papers that cite your root
papers, you can use the option `--also-up`.
Compare for example the two following graphs (root paper in pink):
* Default:
* With `--also-up`:
### Layout and export formats
Export formats are selected using the `--format` (`-f`) option.
The following formats are suitable for importing the graph into
an external graph visualisation tool:
- `gexf`: for [Gephi](https://gephi.org/)
- `dot`: for [Graphviz](https://graphviz.gitlab.io/)
Citegraph can also call Graphviz directly to perform graph
layout and rendering to another format, for example PDF, PNG,
or SVG. The default export format is PDF, see the available ones
with `--help`.
### Customizing graph appearance
You can specify how individual nodes are styled with a yaml file.
For example:
```yaml
tags:
read: # an identifier for the tag
attrs: # DOT attributes: https://graphviz.gitlab.io/doc/info/attrs.html
style: bold
members: # enumerate explicit members using keys of the bibtex file
- someBibKey
- another
knuth_articles: # another tag
attrs:
style: filled
fillcolor: lightyellow
# Select using an arbitrary python expression
# The bibtex entry is in scope as 'paper'
selector: 'any("Knuth" in author.last_names for author in paper.authors)'
```