Cursor movement based user identification, ad targeting and signature verification – interview with Tamás Zelczer CEO at Cursor Insight

Cursor Insight simultaneously develops different solutions, all of which are built on their core movement analysis technology. They can analyse not only our handwriting but also our cursor movements. Their solutions include cursor movement based user identification, ad targeting and user behaviour prediction and verifying digitally recorded handwritten signatures.

ci-logo-with-outlines

 

What is the idea behind Cursor Insight? How did the company story start?

There is a clear link between Cursor Insight’s inception and the fact that a major bank is about to launch a pioneering e-signature solution powered by our technology.

It’s well known that most people have unique, recognisable handwriting styles. We got very excited when we discovered that the fine hand gestures that we use to move a computer mouse are also unique to each of us. Different movement patterns of different individuals can be recognised if we only look at how the computer cursor or pointer is moving around on the screen.

The algorithm that we created can verify the identity of users and even predict some personality traits just by analysing the cursor gestures. This is why we say that your cursor movement is like your emotional signature.

What are the applications of your algorithm and what services does Cursor Insight offer?

We simultaneously develop different solutions, all of which are built on our core movement analysis technology.

Cursor movement based user identification can be used to fight online fraud and increase the security of sensitive services such as online banking applications. We recently took part in a test organised by a leading cyber security company. We were able to verify the registered users of an enterprise software with over 92% accuracy during normal user sessions.

For online advertisers and e-commerce companies we offer the ability to improve ad targeting and predict user behaviour. With our cursor analysis algorithm, we can predict some personality traits such as age, excitement level, or the intuitiveness of anonymous visitors on a website. Our goal here is to create the industry standard of automated personality profiling based on cursor movement analysis.

Example of a user’s cursor dynamics on a website

 

But we don’t only analyse cursor movements. An obvious example for a completely unique human movement pattern is our own signature. Verifying digitally recorded handwritten signatures is currently our most successful product.

Who are your clients?

Automated signature verification is an ideal service for banks since they sign countless contracts and transactions with a large number of customers. We designed and built a secure electronic signature solution to complement our verification capability. Clients can sign electronic documents by only using their handwritten signature, and the digital signature will be just as secure as a traditional chip card based e-signature. Except you don’t need to register for a personal card and buy a card reader.

Banks in Hungary, for example, are required by law to use these very secure, so called advanced e-signatures, where the signatures are undeniable, cannot be replicated and are also protected from theft. Our solution ticks all the boxes and it will be the first to be implemented in the branches of one of the biggest banks in Hungary.

How does the signature verification work?

It works like any other biometric verification system, such as a fingerprint reader. Our algorithm learns what’s unique about a given person’s signature from a few samples. If a registered user signs his or her name again, our system gives a signature match probability score within a second. The underlying technology is extremely complex and we analyse tens of thousands of movement characteristics in the background.

Last year you won SigWIComp2015, an international signature verification competition. How accurate is your verification technology?

We are proud that as a young start up, our algorithm proved to be working better than the competing products of leading companies and university research groups. Of course, a technology like this is never complete. We are continuously working on making it even faster, more accurate and easier to use.

When verifying signatures, we get better results than an experienced human evaluator, such as a bank employee. And of course our artificial intelligence is a lot faster and never gets tired. There are some common signature forging methods that usually trick the naked eye but not our algorithm.

Describe your team! What makes you rocket?

The secret of building an innovative scientific product, like our signature and cursor motion analysis algorithm, is combining the right skills and scientific disciplines. We put a significant amount of advanced mathematics, psychology, physiology, handwriting science and relentless software design and coding into the mix. Our robust products process huge data sets and have to work well in our clients’ complex IT environments. Therefore each of our developers has at least one special power like being an expert in various mathematical frameworks, cryptography, programming in Erlang, Ruby, .NET or being an ethical hacker.

What were the biggest difficulties that you had to cope with so far?

survey-share

We’re in the big data business, which means that we need a large amount of data even in the product development phase. At the beginning we had to be creative when we had no big data provider partners. So we built a web application that includes a personality test. And we then ran an advertising campaign to have enough people fill in our anonymous questionnaire and move around with the cursor in the process so we can feed data into our algorithm. The test is accessible from our website so feel free to give it a try and get a personality feedback in 5 minutes.

Please follow and like us: