https://github.com/chainsawriot/totervogel

⚰️🐦 Toter Vogel (Dead Bird): A Simple Digital Forensic Tool for Twitter

https://github.com/chainsawriot/totervogel

Repository

⚰️🐦 Toter Vogel (Dead Bird): A Simple Digital Forensic Tool for Twitter

README.Rmd

---
output: github_document
---



```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
```

# totervogel 




The goal of totervogel is to detect malicious twitter accounts using Benford's Law.

Please refer to Golbeck (2015) and Golbeck (2019) for more details about the analysis of first significant digits. For the analysis of last significant digits, please refer to Dlugosz & Müller-Funk (2009)

1. Golbeck, J. (2015). Benford’s law applies to online social networks. PloS one, 10(8), e0135169. doi: [10.1371/journal.pone.0135169](https://doi.org/10.1371/journal.pone.0135169)
2. Golbeck, J. (2019). Benford’s Law can detect malicious social bots. First Monday. doi: [10.5210/fm.v24i8.10163](https://doi.org/10.5210/fm.v24i8.10163)
3. Dlugosz, S., & Müller-Funk, U. (2009). The value of the last digit: Statistical fraud detection with digit analysis. Advances in data analysis and classification, 3(3), 281. doi: [10.1007/s11634-009-0048-5](https://doi.org/10.1007/s11634-009-0048-5)


## Installation

You can install the development version of totervogel from [GitHub](https://github.com/) with:

``` r
# install.packages("devtools")
devtools::install_github("chainsawriot/totervogel")
```

## Example

This is how the totervogel of an organic human account looks like. By default, it analyzes the friends.

```{r myaccount}
library(totervogel)
res <- create_totervogel("scott_althaus")
res
```

```{r plotmyaccount}
plot(res)
```

You can also analyze followers.

```{r followers}
res_fol <- create_totervogel("scott_althaus", followers = TRUE)
res_fol
```

```{r folplot}
plot(res_fol)
```

A potentially malicious twitter account's totervogel results might look like:

(Please don't visit these accounts.)

```{r malicious}
malicious_res <- create_totervogel("badluck_jones")
malicious_res
```


```{r malicious2}
malicious_res2 <- create_totervogel("yoyo13148779", followers = TRUE)
malicious_res2
```

```{r plotmalicious2, echo = TRUE}
plot(malicious_res2)
```

# Notes

* In Golbeck (2015), 89.7% of Twitter users had a correction of over 0.9. Less than 1% had a correlation under 0.5. An account must be very suspicious to have such a low correction.

* Accounts with a lower friends count are more likely to be detected with lower Benfordness.

* The last digit analysis is experimental. The results should only raise your eyebrows, if more than one aspect (friends, statuses, followers) displays unexpected distribution.

* The logo of this package is a remix of Kearney et al's  [rtweet](https://github.com/ropensci/rtweet)'s logo. The original logo is licensed under an MIT License.

Owner

GitHub Events

Issues and Pull Requests

Last synced: over 1 year ago