Political Information, New Media, and Twitter bots

class: center, middle, inverse, title-slide

# Political Information, New Media, and Twitter bots
### Michael W. Kearney📊 School of Journalism Informatics Institute University of Missouri
### <table style="border-style:none; padding: 0 0 50px 0; background-color:transparent" class="table">
<tr>
<th style="padding:0 30px">
<a href="https://twitter.com/kearneymw"></a>
</th>
<th style="padding:0 30px">
<a href="https://github.com/mkearney"> </a>
</th>
</tr>
<tr style="background-color:transparent">
<th style="padding:0 30px">
<a href="https://twitter.com/kearneymw"> @kearneymw </a>
</th>
<th style="padding:0 30px">
<a href="https://github.com/mkearney"> @mkearney </a>
</th>
</tr>
</table>

---

class: center, middle

Presentation created using **{xaringan}** (the **robot** theme)

Slides available at [mkearney.github.io/dsa_execweek_talk/](https://mkearney.github.io/dsa_execweek_talk/)

---

background-size: 150px auto
background-position: 490px 185px, 567px 320px, 644px 185px, 721px 320px, 644px 455px, 567px 50px
background-image: url(img/chr-logo.png), url(img/hexagon-logo.png), url(img/textfeatures-logo.png), url(img/tfse-logo.png), url(img/botrnot-logo.png), url(img/rtweet-logo.svg)

# About Me

**Background**

- PhD in COMS from Kansas

**Research interests**

- Partisan selective exposure
- Digital and social media

**\#rstats packages**

- Twitter APIs: [**{rtweet}**](https://cran.r-project.org/package=rtweet)
- Text analysis: [**{textfeatures}**](https://cran.r-project.org/package=textfeatures) [**{chr}**](https://github.com/mkearney/chr/)
- Data wrangling/viz: [**{hexagon}**](https://github.com/mkearney/hexagon/) [**{tfse}**](https://github.com/mkearney/tfse/)
- Machine learning: [**{botrnot}**](https://github.com/mkearney/botrnot/)

**Shiny apps**

- Web app interface for [botrnot](https://mikewk.shinyapps.io/botornot/)
- [Interactive friendship tool](https://mikewk.shinyapps.io/friendship/)

---

# Overview
+ The **problem** and the **research question**
   - Use of bots on social media to manipulate public opinions, amplify social divisions, and spread "fake news"
   - **How do we accurately and dynamically classify bots on social media?**

+ Proposed **solution**
   - A new, fast, and flexible approach to training models to detect social media bots

+ **Future** directions
   - Things to consider moving forward

---

# 2016 election

Concerns about **automated accounts, or bots, on social media** manipulating public opinion reached a fever pitch during the 2016 general election.

The most alarming concerns were related to behaviors of **Kremlin-linked bots** for:

+ pushing **fake news** stories

+ amplifying **social divisions**

+ misrepresenting **public opinion**

---

# Research question(s)

## How do we **identify bots on social media**?

## How can we do it in **real-time** so as to filter out inauthentic social media traffic?

---
class: inverse, center, middle

# Social media data

---

# Social media apps

There's been a lot of talk about Cambridge Analytica's abuse of social media data. **Reactions** can be summarized as follows:

1. **Facebook was negligent** because it inadequately regulated third party apps

1. **Cambridge Analytica was unethical** because it stole and used data they shouldn't have

&nbsp;

In other words...people don't know exactly what FB and CA actually did, but they know it was wrong.

---

# Social media APIs

Social media platforms encourage and regulate third party applications via **APIs**
+ **Application Program Interfaces** (APIs) are sets of protocols that govern interactions between sites and users

Similar to web browsers but with different primary objective:
+ Web browsers **render content**
+ Web APIs **organize data**

For public APIs, many sites only allow access to **authorized** users
+ Twitter, Facebook, Instagram, Github, etc.

---

# Twitter vs. Facebook

For a case study about political bots on social media, Twitter is better because...

1. It's more open about its **data**
   + Unlike Facebook, which actually clamped down on data sharing (with developers) starting in 2014, **Twitter is quite open with its data**

1. It's more **public facing**
   + **Default privacy setting** is public
   + User **connections** are asymmetrical

---
class: inverse, center, middle

# rtweet

---

# Twitter APIs

To connect to Twitter APIs, users need a **consumer key** and **consumer secret**, which you can get by creating a Twitter app.

View the built-in vignette **`auth`** for instructions on obtaining and storing user access tokens.\*

```r
## view rtweet's authorization vignette
vignette("auth", package = "rtweet")
```

Or look up the vignette online: [rtweet.info/articles/auth.html](http://rtweet.info/articles/auth.html)

.footnote[
 \* For a while, rtweet users were not required to create their own Twitter applications. [That recently changed](https://github.com/mkearney/rtweet/issues/179#issuecomment-365689068), but hopefully it's resolved soon.
]

---

# Create a Twitter App

1. Log in to your [Twitter.com](https://twitter.com) account

2. Go to [apps.twitter.com](https://apps.twitter.com/) and select **Create New App**

---

3. Complete the fields (**important**: Callback URL must be exact)
 - Name: `{{initials}}_twitter_app`
 - Description: `{{something about analyzing Twitter data}}`
 - Website: `https://twitter.com/{{you_screen_name}}`
 - **Callback URL:** `http://127.0.0.1:1410`

4. Store name of app as `app_name` in R script

```r
## name of twitter app
app_name <- "mwk_twitter_app"
```

---

# Copy API keys

5. Click **Create your Twitter application**

6. Select **Keys and Access Tokens** tab

7. Copy the **Consumer Key (API Key)** and **Consumer Secret (API Secret)** and paste into R script

```r
## copy and pasted *your* keys (these are fake)
consumer_key <- "XYznzPFOFZR2a39FwWKN1Jp41"
consumer_secret <- "CtkGEWmSevZqJuKl6HHrBxbCybxI1xGLqrD5ynPd9jG0SoHZbD"
```

---

# Create your token

8. Pass the stored values to `create_token()` to create the Twitter authorization token.

```r
## create token
token <- create_token(app_name, consumer_key, consumer_secret)

## print token
token
```

---

# Storing your token

Every request sent to Twitter must include a token. Save yourself trouble by storing it as an environment variable.

```r
## save token to home directory
path_to_token <- file.path(path.expand("~"), ".twitter_token.rds")
saveRDS(token, path_to_token)

## create env variable TWITTER_PAT (with path to saved token)
env_var <- paste0("TWITTER_PAT=", path_to_token)

## save as .Renviron file (or append if the file already exists)
cat(env_var, file = file.path(path.expand("~"), ".Renviron"),
  fill = TRUE, append = TRUE)
```

Normally `.Renviron` is processed on start-up, so refresh it.

```r
## refresh .Renviron variables
readRenviron("~/.Renviron")
```

---

# Getting the data

**{rtweet}** functions correspond to different [API] paths. For example, user timelines:

```r
cnn <- get_timeline("cnn", n = 3200)
ts_plot(cnn, "hours")
```

---
class: inverse, center, middle

# Identifying bots

---

# Current approach

The creators of [**botometer**](https://botometer.iuni.iu.edu/#!/) maintain a list open sourced academic studies that identified bots.

But this approach is not without **limitations**:
+ Academic research moves slowly (especially with Twitter banning many of the bots)
+ Relatively small number have been collected
+ Labor intensive methods (human coding)

---

# New approach

One **solution** to these limitations is to combine the academic sources with an easy-to-automate method that takes advantage of naturally occurring human coding

1. Select a handful of previously **verified** bots
1. Look up the public **lists** that include those bots
1. Identify list **names** that self identify as bot lists
   - Perform validity checks on other listed accounts

---
class: inverse, center, middle

# Compiling data sets

---

# User/tweet data

Using the list of screen names for non-bot (based on previous academic research and human coding) and bot accounts, I gathered...

+ **tweet-level data** for the most recent 100 tweets posted by each user `rtweet::get_timelines()`

+ **user-level (or account-level) data** for each user `rtweet::lookup_users()`

The data returned by Twitter consists of over 60 features related to the tweet and author (user) of each tweet.

---

# Data sets

Randomly sample Twitter data into train and test data sets

+ **Train data**: approx. 70% of users in sample
   - Making sure there are a roughly equal number of bots and non-bots

+ **Test data**: remaining (30%) bot and non-bot accounts

---

# Features

In addition to the numeric variables already returned by Twitter, I used **{textfeatures}** to extract up to 26 features for each textual variable.

The text features were extracted for:

+ Tweets
+ Name
+ Screen name
+ Location
+ Description (bio)

---
class: inverse, center, middle

# Modeling

---

# Two models

Two models were created.

1. The **default model**, which uses both users and tweets data.
   - Rate limited by Twitter to 180 every 15 minutes
   
1. The **fast model**, which uses only users data
   - Rate limited by Twitter to 90,000 every 15 minutes

---

# Model results

A gradient boosted logistic model **{gbm::gbm()}** was trained and tested on the two data sets for both the default and fast models.

+ The **default model** was 93.53% accurate when classifying bots and 95.32% accurate when classifying non-bots

+ The **fast model** was 91.78% accurate when classifying bots and 92.61% accurate when classifying non-bots

Overall...

+ The **default model** was correct 93.8% of the time

+ The **fast model** was correct 91.9% of the time.

---
class: inverse, center, middle

# botrnot

---

# Applications

+ This has since been assembled into an R package [**{botrnot}**](https://github.com/mkearney/botrnot)---coming soon to CRAN!

+ It has also been exported as a [Shiny web app](https://mikewk.shinyapps.io/botornot)

---
class: inverse, center, middle

# Future directions

---

# Future research

Between the proliferation of news stories about Kremlin-linked bots on social media and the number of false positives reported to me about **botrnot**, I'm starting to think...

The real challenge is not **identifying** social media bots but instead **defining** what is means to be a *bot* on social media

+ Future research should examine how and why people define certain accounts as bots and others as not bots

---
class: inverse, center, middle

# The end