Let’s talk about spam

We do our best to provide users of LOVOO with a great experience. Combating spam is a key component of this. To us, it is important that users have confidence in us and our product. We now want to boost this confidence with more transparency.

Last year we reported extensively on the technical implementation of anti-spam solutions at LOVOO. Combating false and harmful user profiles still plays a crucial role in day-to-day operations and many teams are needed for this. This article will take a closer look at what this actually involves.

Figure 1: LOVOO’s Anti-Spam Team works behind this door

The anti-scam system of LOVOO is specialized in the detection and combating of spam, scam and fake profiles using automated and semi-automated processes. How do these individual profile types differ?

Spam
Spam attacks often involve placing dishonest advertisements for other products and enticing users to external websites or apps. This normally occurs via chat messages or advertising text in a profile with reference to other platforms. Text such as “I am going to delete my profile, you can find me at eine-andere-plattform.net/susi1234” is particularly popular. Advertising messages embedded in profile photos are also often used.

Der Spam-Friedhof unseres Lab-Teams.

Figure 2: The spam graveyard of our Lab Team.

Scam
Contrary to spam, so-called scam profiles can cause significant harm. Distributors of these profiles are consciously searching for users that they can entice with bait advertising and later blackmail.

Fake
Fake profiles mimic an identity other than that actually behind them. Fortunately, these are few and far between and are usually detected and stopped by other users and customer support within a very short space of time. Attackers also rely on fake profiles to distribute spam and scam content.

Is there an “anti-spam alert system”?

For the sake of simplicity, I will collectively refer to the three types of profiles – spam, scam and fake – using the term “spam” in the following.

The challenge of combating spam sustainably over the long term does not only concern LOVOO, but the entire industry. Social networks such as Facebook, Instagram or Snapchat are regularly victims of spam attacks.

It would appear that, with such a wide market penetration of this problem, there is a uniform technical solution. This is unfortunately not the case. A proactive and intelligent system is required to prevent attackers from succeeding in penetrating the controls of a platform. Learning the algorithms and processes is a fundamental requirement for continued development and keeping spam at bay as far as possible.

Verzögert identifizierte Spam-Profile werden nach abgeschlossener Analyse manuell durch das Anti-Spam-Team blockiert.

Figure 3: Spam profiles identified late are manually blocked by the Anti-Spam Team once an analysis has been completed

Spam profiles identified late are manually blocked by the Anti-Spam Team once an analysis has been completed. Information on these attacks is applied in training, and therefore in the continued development of real-time anti-spam algorithms, in order to be able to respond more quickly and ward off recurring attacks. Repeated waves of attack therefore have no chance of success.

Here is a brief explanation of the technologies and processes applied by us in order to give you a comprehensive overview of what LOVOO does to prevent spam:

Prevention

  • Encrypted and securely authenticated data transmission between the LOVOO app or website and the LOVOO system makes it more difficult for attackers to falsify or misuse data transmissions.
  • Invalid or duplicate email addresses, which could be used for the creation of spam profiles, are prevented.
  • Users who have verified their profile are identified as such so that every user can clearly recognize whether a user is genuine or not.
  • Profiles with limited verification (e.g. no valid email or photo verification) will have limited access to the LOVOO platform.
  • The intelligent and rule-based algorithms of the anti-spam system are continuously improved on the basis of previously detected and repelled spam attacks.

Identification

  • Spam profiles are automatically identified in real time with rule-based and intelligent machine learning algorithms.
  • Warnings from users via the LOVOO app, website or via email are evaluated manually and automatically in order to identify both individual spam attempts and bigger spam waves.
  • New information regarding the behavior of potential spammers is used to subsequently search user profiles for spammers and identify any that may be present.

Combating

  • Identified spam profiles are blocked manually or automatically.
  • Blocked users may only communicate with others again once they have verified their profile (e.g. with photo or SMS verification). This also ensures that so-called false positives (profiles incorrectly classified as spam) may also continue to use their profile.

The authenticity of profiles

LOVOO offers various verification methods that ensure that a user is the person that he or she claims to be. These are difficult or impossible to automate and therefore require a considerable investment of time in the pages of users. They give a profile a higher level of importance and are particularly suited to prompting profiles classified as potential spam to verify their authenticity.

LOVOO offers the following verification methods:

  • Email verification (optional): Ensures that users use authentic email addresses and cannot create countless fake profiles.
  • Facebook login (optional): Ensures that the logged in user also has a valid Facebook profile. Bringing Facebook to the table also makes spam protection effective.
  • Photo verification (optional): Ensures that the user shown in the profile photos is actually the person behind the profile.
  • SMS verification (mandatory for profiles classified as spam): Ensures that the user of the profile has a cell phone number and actually exists. This prevents the automatic creation of profiles on the LOVOO platform.

Furthermore, blocked and inactive profiles receive an automated request to re-verify the profile following a certain period of time. If this is not completed within seven days, the profile will be permanently removed from the platform. This ensures that only users that have been active in the recent past will be displayed.

Why women in particular are often reported

We always try to be a step ahead of spammers that attack LOVOO. However, a completely spam-free LOVOO is currently not possible. We receive thousands of reports from users regarding potential spam profiles via the app and email on a daily basis. The most common reports are initially examined by our customer support team.

It is frequently the case that the reported profiles involve genuine users that have simply not responded to contact attempts. This phenomenon is particularly observed when men write to women and do not receive a response.

The writers of the chat requests often assume that it must be a fake profile and subsequently report it. In more than 85% of cases, the women are simply not interested or dislike the way the chat request was written and therefore reject it. This is at least in part due to vulgar or direct formulations.

More updated spam reports

In Arbeit: Der LOVOO Fake & Spam Transparency Report

Figure 4: Currently in doing: the LOVOO Fake & Spam Transparency Report

In order to create more transparency, we will publish regular reports on spam attacks on LOVOO and the tackling of these with our anti-spam algorithms in future. An initial report on this with interesting and enlightening figures will follow in a few weeks.