Brand Analytics, a resident of the Skolkovo Foundation, has prepared and launched the Russian Fault Detector service, which already at the start shows a picture of the performance of more than 100 popular Internet services. In the observed pool, mobile applications of banks and delivery services, online stores, transport infrastructure, etc.
Until recently, this problem was solved by the foreign service DownDetector (DD). However, in the spring of 2022, he stopped working in Russia.
Oksana Ulyaninkova, the head of promising projects in the field of information security of the information technology cluster of the Skolkovo Foundation, said: “The main advantage of this service is its focus on the Russian market, that is, the specifics are taken into account and specific resources of our country are monitored. The solution will be in demand by both users and partners who will be able to use data and analytics in their work. If DownDetector had not left Russia, Fault Detector would have taken the lion’s share of the local market anyway.”
Natalia Sokolova, CEO of Brand Analytics, said: “Today, companies are more likely to learn about failures in their services from users than from internal services. People share their emotions on social networks due to problems with the service, counting on the support and advice of friends. However, in order to find such information and understand what is happening, it is necessary to analyze a very large data stream and use real-time neural networks for this. Our failure detector can do this and will be able to alert companies faster than other sources when they have problems.”
The failure detector is not an analogue of the DownDetector (DD), it is a step or even two steps forward, they say in Brand Analytics. First, DD could only analyze Twitter. With a 34% drop in the number of active social media contributors from February to May 2022 according to Brand Analytics research, using only that social media for a service like Detector is a controversial decision.
Unlike DD, Crash Detector analyzes all sources of social media and, first of all, the largest ones – the Russian social networks Vkontakte and Odnoklassniki, as well as Telegram.
Moreover, many of the popular services are partners of Brand Analytics and use the analytics system of the same name in their work. The technology for identifying failures has been working inside the Brand Analytics system for several years and has proven its effectiveness in practice. Now it has formed the basis of a public service.
Further development of the “Detector” will without fail take into account feedback from both partners and directly from users of social networks.
To identify problems in the operation of various services, the Detector neural network analyzes over 40 million Russian-language user messages in various social media sources every day and finds information about failures.
The detector represents a set of dashboards, each of which relates to a separate service. The dashboard of each service shows its status (available, there is a possibility of failure, failure), the dynamics of failures “through the eyes of users”, the identified problems, the centers of failures on the heat map of the world and the message feed, which helps to understand the situation.
The presented version of the service is the first, but not final. Brand Analytics plans to further develop Detector, taking into account feedback from partner companies, as well as from users.