Everyone is embracing user generated content – from mainstream newspapers like the Guardian and New York times, to social media (blogs, etc) and all the way to enterprise 2.0 – McDonalds, Marriott, BP and others, encouraging people to talk about their experiences with products, reviewing their latest vacation etc. But this level of freedom poses a big threat to advertisers. What tools are available today to prevent the advertised brands from being trashed right next to the ad?
Until now, this problem has created a big headache to the social media providers, who had to deal with moderating long queues of user generated content. Israeli startup Contrust (“Content you Can Trust”) developed technology to solve exactly that pain. They serve as a filter, deciding if the content is good or bad for the site.
I learned more about Contrust from co-founder and CEO Shai Wolkomir, a young ‘veteran’ of the security industry. Shai was a team leader at Checkpoint and a former product manager at RSA. I previously covered Contrust’s funding when they were known as Celtego (see The land of milk and honey: Ten Israeli companies announce rounds in less than two weeks). Contrust recently debuted in TechCrunch50, and it is entering a space where Keibi Technologies, Mollom and Automattic are trying to establish dominance.
Contrust is in closed Beta, but Shai agreed to share a special invitation with the readers of VC Cafe and keep you on the loop. Sign up through this link.
VC Cafe: Hi Shai. May I ask the obvious question? What exactly is appropriate? (or are we becoming China?)
Shai: We enable a baseline. First we protect against everything that is taboo (profanity, violence, racism). Since we sit in the upload stream – before anyone published the content, so we stop it from even getting served. Part of our moderation elements are not contextual. We can also identify abusers on a user level, not only using the classic content analysis.
VC Cafe: What threats is Contrust able to stop?
Shai: You have to understand the motivation of the bad guys to stop them. We detect spam not according to the content. There are two type of spams – bot spam and manual spam. In user generated content sites, the spam is done for the purpose of SEO and less for driving clicks.
VC Cafe: How do you stop the spam from being published without hurtuing the publisher?
Shai: Contrust serves as a trusted pipe. We can understand if you’re human or bot – by measuring how long did it take you to write the post, for example. Less than 3 seconds for 400 words? Did you use copy paste? Did you land on the submission page directly? All these clues makes your entry more suspicious.
VC Cafe: Can you explain a little about the backend technology?
Shai: Sure. Contrust is sitting in the cloud – we are a classic web service XML. Before an entry is published, we retrieve it and assign it a risk score. The clients don’t want to mess with any old school sdks, or complicated installs. They want Saas. We use behavioral recognition techniques to identify threats. We can see many types of content in industries.
Our system is a learning one. There is a full management system. Companies have terrible content queue management. We offer the moderators a system that categorizes the threats. We also build the profile for a user each part of the site. The process can never be fully automated, but instead of paying 50 people to sitin the room, we simplify the process by cleaning up most of the queue for them. Spam could easily be automated. Identifying taboo is easy.
VC Cafe: What are Contrust’s biggest challenges?
Shai: One of the amazing things about our startup is the ‘meet the market’ – we are trying to sell moderation, to someone who doesn’t want to hear about security or filtering. We are also trying to bring a content management system to a world that is usually much simpler. That said, we’re being accepted with amazing enthusiasm.
VC Cafe: Let’s get back to some of the basics. When did you start and how are you funded?
Shai: We’ve been running informally since February 07. We were Incorporated in Nov 2007 and raised $700,000 in a round led by Xenia Capital and the founders of Applied Semantics (The Elbaz family). We’re currently 7 people, including outsourcing. Our Beta was launched a month ago.
These days we’re getting to operate new clients.
VC Cafe: Are you planning to establish a precense in the US?
Shai: I went to the US for 2 weeks and ended up staying for a month because of follow up meeting requests. I understood how urgently we need to have presence in the US, because that’s where our market is. So we’re thinking about raising an intermediate round, to create a US presence as well as getting enough fuel until 2010.
VC Cafe: Every site has different types of content and different filtering. How do you customize the platform to the needs of each individual client?
Shai: Clients can set up channels, according to their own configuration. Anti porn, anti spam, anti phishing. Each entry is getting a score. You can define that messages that are 100% pass automatically, and messages that are 0-90% go to moderation.
Contrust gives a recommendation to approve or decline. Content reputation systems show the moderator what other moderators did in similar messages. For example, there’s a post that doesn’t have anything suspicious but includes a link to a porn site. The system lets you preview the page on the CRM system. The users can also tag words inside the message and ‘mark as bad’, as opposed to the old systems that include a queue of messages that have no tagging.
VC Cafe: Ho defensible is your technology? Do you have any patents?
Shai: We pass information on the user history (could add content on the user using cookies) and we are working on a technology to identify people in a more efficient way that doesn’t require cookies.
Israel has been known as a security leader in the enterprise world, but the world of web 2.0 is very different. Users expect to publish the content they contributed immediately, and the expectation is for zero moderation. Plus, the threats are new. I was impressed with Contrust’s solution and I’m sure we’ll be hearing more about them in the future. Keep in touch with Contrust through their blog and sign up for the closed beta in VC Cafe’s page on Contrust.
Latest posts by Eze Vidra (see all)
- Emerging Machine Intelligence Clusters - March 2, 2017
- 30 Machine Intelligence Startups to Watch in Israel - February 15, 2017
- Five Books I Want to Read in 2017 - January 16, 2017