Overview

In this job, you are asked to review and classify Twitter accounts in order to assess whether they are either Spambots or Genuine accounts. A spambot is an account that tweets in an automated way, usually with the goal of promoting a product/brand, or of publicizing certain events and/or some target (e.g., a video or song, a politician, a job offer, etc.). The information gathered with your precious help will be used to train a software for the automatic detection of spambots on Twitter.

What we Provide

For every Twitter account we provide the following information:

• Account unique numeric identifier (Account ID);
• Account Name;
• Account Screen Name;
• Link to the account page on Twitter.

Process

1. Read the account name and screen name.
2. Click on the link to the account page on Twitter.
3. Verify that the account name and screen name correspond to that reported in the account page on Twitter.
4. Choose "Unable to classify" if you are unable to correctly see the account page on Twitter, or if the information in the account page are different with respect to those reported in the job interface. Among the possible problems that you might encounter are: (i) deleted accounts; (ii) suspended accounts; (iii) protected accounts, which are those accounts whose tweets are private. Please, add a brief description of the reasons that prevented you from classifying the account.
5. If the account page displays correctly, take your time to browse and explore all the relevant information in the page: name, photo, biography, friends & followers, recent tweets, uploaded pictures & videos, any other available information.
6. When you have made up your mind about the account, close the account page and choose one of the 2 following categories:
   • "Spambot";
   • "Genuine".
7. Repeat these steps for the next account.

Rules and Tips

Do This

Unfortunately, there is not a universal definition of a Twitter spambot. However, Twitter spam is somewhat similar to email spam, and there are a number of signs that can help you to detect a spambot when you see it. For example, look for the following signs of spambot activity:

• the account does not seem to belong to a real person;
• it tweets the same tweet multiple times;
• it frequently tweets about ways to gain followers;
• it repeatedly asks for followers/retweets;
• it frequently tweets about sex/porn/dates;
• it frequently tweets about ways to make money/job offers;
• it frequently tweets about products on sale;
• it repeatedly offers money/cash/promotions/discounts/prizes;
• it tweets many URLs, possibly all pointing to the same domain;
• it promotes a specific hashtag/URL/product/brand/job position/etc.

Anyway, don't be afraid to use your instinct! Just remember: if the account does not show clear signs of automation/spam, it should be classified as "Genuine".

Take advantage of all the information possibly available on the account page, such as:

• profile image (photo);
• background image;
• name and screen name;
• short biography;
• location;
• web site;
• number of tweets;
• number of followers & friends;
• the list of followers & friends;
• and of course... the content of tweets!

Be Careful

Some accounts might not be tweeting in English or in your native language. Anyway, you should be able to assess the nature of the account (spambot or genuine) without the need to understand its tweets, by focusing on all the signs and the information described above. You can also resort to an online translator, such as Google Translate, if you feel like translating the tweets.

Some accounts might have shared just a few tweets. This might make it difficult to classify them. Anyway, try your best to exploit those few tweets and all other available information (photo, description, friends & followers, etc.) to take a decision. If the available information does not provide convincing evidence of spambot activity, simply classify the account as "Genuine".

You might want to check "Tweets & replies" and "Photos & videos" in order to better evaluate an account (see the image below).

Do Not

Do not focus exclusively on the most recent tweets. Instead, try scrolling down in the account timeline in order to see its older tweets.

Do not consider "Promoted tweets". These are advertisement tweets injected by Twitter and not tweets produced by the account, simply ignore them (see the image below).

Do not use "Unable to classify" if the account has little information. Only use "Unable to classify" when a problem occurs, such as for deleted, suspended and protected accounts. In all those cases when you are not sure about spambot activity, simply classify the account as "Genuine".

Examples

Example #1

Account ID: 14596967
Account Name: fxgenie
Account Screen Name: @fxgenie
Link to the account page on Twitter: www.twitter.com/fxgenie

Decision: Spambot

Reasons:

The main reason for classifying this account as a "Spambot" is that it always tweeted the same (or slightly modified) message. These tweets have the goal of making other users click on the links that are added at the end of every message.
Please, note the presence of a promoted tweet in the screenshot on the right. This tweet should be ignored since it is an advertisement tweet injected by Twitter.

Example #2

Account ID: 44705716
Account Name: USGSted
Account Screen Name: @USGSted
Link to the account page on Twitter: www.twitter.com/USGSted

Decision: Spambot

Reasons:

Although this account does not perform any malicious activity, it is clearly fully automated and should be classified as "Spambot". All the tweets have the same predefined "structure". Please, also note the lack of interactions with other accounts, and the lack of uploaded photos or videos. Also, the account does not follow almost any other account, as represented by the extremely low number of followings (friends).

Example #3

Account ID: 66086468
Account Name: Eliana Young
Account Screen Name: @eliblack996
Link to the account page on Twitter: www.twitter.com/eliblack996

Decision: Spambot

Reasons:

Although this account seems to belong to real person (note the credible name and profile picture) it is nonetheless a Spambot. The limited information available for this account, only 5 tweets, still allows to notice the intention to lure users into clicking the link included in every tweet. The profile photo of a young and attractive woman, and the provoking content of tweets, add to the evidence of spambot activity.

Example #4

Account ID: 27159435
Account Name: bigorna
Account Screen Name: @_bigorna
Link to the account page on Twitter: www.twitter.com/_bigorna

Decision: Spambot

Reasons:

An added difficulty for the classification of this account is that it tweets in a language other than English. Anyway, despite not being able to thoroughly understand the content of tweets, a detailed analysis of the account still allows to clearly assess its spambot nature.
On the one hand, the general information about this account does not seem very convincing: the profile picture is a generic photo of an anvil, the account has no description, location, url, etc. On the other hand, the 3 most recent tweets are retweets to other users' tweets. Interactions with other users are typically signs of human activity. However, the 3 retweeted tweets all have the exact same content. Moreover, those tweets talk about ways to get more retweets. An analysis of all the other tweets shared by this account allows to notice that it frequently refers to "seguidores", which means "followers".
Typically, frequent references to getting more retweets and more followers are a strong sign of spam. In addition, almost every tweet has a url, and many tweets have the hashtag #MaisFollowers. For all these reasons, this account provides sufficient evidence of spambot activity.

Example #5

Account ID: 150272164
Account Name: Janet
Account Screen Name: @Bieber_rose
Link to the account page on Twitter: www.twitter.com/Bieber_rose

Decision: Genuine

Reasons:

This account presents detailed and credible general information, such as a personal photo, description, location, url, birthday, etc. It also has many interactions with other accounts in the form of retweets and mentions. Also, it uploads photos and shares links to news and music videos. This account also features a very high number of followers and friends, the majority of which seem to be genuine (human-driven) accounts.
All these signs clearly support the classification of this account as Genuine.

Example #6

Account ID: 117712677
Account Name: VoltimumUK
Account Screen Name: @voltimumuk
Link to the account page on Twitter: www.twitter.com/voltimumuk

Decision: Genuine

Reasons:

This account probably refers to the official Twitter account of the self-declared "leading portal for the Electrical Industry". This means that it does not refer to a real existing person. Anyway, this is not enough to claim that this account is fully automated, or that it "spams" tweets. Indeed, despite having links pointing to the firm's Web site, all tweets are rather different from one another and seem to lack a predefined structure. In the end, it's not possible to tell whether the account is automated or human-operated. As such, it should be classified as Genuine.

Example #7

Account ID: 3520772356
Account Name: bibliotecagbnicolosi
Account Screen Name: @bibliopaterno
Link to the account page on Twitter: www.twitter.com/bibliopaterno

Decision: Genuine

Reasons:

This account presents very limited information to help us with the classification: no personal photo as profile image, a seemingly non-human name, no other personal information, just 2 tweets (moreover, not even in English), and a bunch of followers. In this situation, we do not have enough evidence to say that the account is a "Spambot", and we must classify it as "Genuine".

Example #8

Account ID: -
Account Name: -
Account Screen Name: -
Link to the account page on Twitter: -

Decision: Unable to classify

Reasons:

Whenever a problem prevents you from correctly accessing an account's page on Twitter, choose "Unable to classify" and add a brief description of the problem. As shown on the right, among these problems are those cases when an account has been suspended by Twitter, when a user intentionally deleted his/her account, and when a user chose to make his/her tweets private (this is called a protected account).