The robot follows the news
ABBYY has implemented a unique online news monitoring system in Sberbank. Built on artificial intelligence technologies, the solution automatically analyzes the content of messages about more than a thousand competing banks, allowing the client to quickly track important information on the market. & #13;
It is especially important to keep abreast of the information field in the financial sphere. To notice a new product or service of a competitor in time means to be able to quickly join the competition and defend your market positions.
Humanity generates so much information that it would take a whole army of specialists to collect and analyze it “manually”. Even if we exclude “noise” and sources that are not related to business from this data flow. However, for such a giant as Sberbank, everything that happens in the global economy, every movement of the global and local market is important. This information is needed to make decisions about investing and selling assets at the moment when it is most profitable. It is especially important to observe how the competitors are doing. The existing systems for collecting news from public sources do not allow automating the process. Employees spend a lot of time on analysis.
Solution: the algorithm analyzes the content
You can quickly scan and analyze the information space online news monitoring systems. Their developers have recently been actively using artificial intelligence technologies that allow them to automatically track news that is significant for a particular company.
ABBYY, one of the world’s leading developers of solutions in the field of intelligent information processing and linguistics, has recently implemented in Sberbank a unique system for online monitoring of news about competitors. “The uniqueness of the project is that natural language processing technologies were used for automatic analysis of unstructured texts, – told Digital Enterprise Olga Morozova, Director of Financial Sector at ABBYY Russia. — They help to sort the news by its essence without human intervention and to identify with high accuracy whether significant names, names of organizations and departments, addresses are contained in messages about banks, to determine the relationships between them. Unlike ABBYY technologies, many media monitoring systems on the Russian market use other data processing tools rather than semantic analysis. For example, some solutions provide a search for individual words and filters for positive and negative events. This makes it easier to work with information, but does not allow you to automate the process. Employees need more time to find and highlight meaningful data in the general news flow.”
What types of news can an intelligent information monitoring system quickly track? It depends on the customer. For example, large financial organizations include news about financial instability, the change of ownership of a major player in the market, the bankruptcy of a significant enterprise as sources of risk. The developers of such systems emphasize that the most difficult thing in the implementation of such projects is to choose the right uniform requirements for the extraction of disparate data. And this, in turn, is necessary in order to track most messages with risk factors and correctly determine the type of risk. “In the development of solutions, we always take into account the interests of customers, — explains Olga Morozova. — For example, for the banking sector, the analysis of news about counterparties in various sectors of the economy is relevant, the processing of messages not only in Russian, but also in English and other languages. In addition, ABBYY technologies can be scaled to other banking business processes — for example, for monitoring and calculating investment, credit, currency and other types of risks.”
Result: not only for banks
According to the developers, the new ABBYY solution works as follows: the bank’s database receives several thousand news about counterparties every day. A specially created algorithm analyzes the content of these texts, finds the so-called risk factors important for the customer in them and arranges messages according to the type of risk. After the artificial intelligence identifies significant messages, the system extracts objects from them and determines the connections between them: the type of event, the name of the organization, positions, and so on. Then this structured information enters the internal system, where an employee of the bank looks at it and makes a decision on the rating of certain operational data. “With the help of AI, the bank monitors such messages in real time. If earlier the bank’s employees processed hundreds of news items per day, now the bill goes to thousands. Thanks to ABBYY technologies, employees save time on searching for information, and the released resources can be used in other areas of credit analysis”, — says Olga Morozova.
Scheme of operation of the ABBYY online news monitoring system
Representatives of Sberbank indicate that they are satisfied with the first results of the introduction of an intelligent news analysis system and in the near future the company intends to further improve this solution. “After studying the world practice, we came to the conclusion that there are few similar projects in the banking sector. It was decided to develop the project further, ” says Dzhangir Dzhangirov, Senior Vice President of Sberbank. According to him, in the near future, the bank plans to integrate with a foreign news provider to launch monitoring of the news flow in English, test and refine the implemented models, as well as create a classification of news by economic sectors. Participants of the IT market are convinced that in general the demand for intelligent solutions for information monitoring will grow, and not only from banks, but also in large companies in other industries. “Business needs to respond quickly to significant information events, as negative news can affect the reputation and financial stability of the organization. The number of messages is already estimated in the hundreds of thousands per day. It is difficult to have time to read and analyze such a huge flow of information without artificial intelligence and natural language processing technologies”, – he says Olga Morozova.
According to ABBYY, employees of large companies spend at least 1 hour a day, or more than 240 hours a year, to find all messages that are significant for business, check information and prepare analytics. The technologies of intelligent search cope with this task approximately 3 times faster. In addition, they track information better: according to the results of the system, the robotic solution finds 85% or more relevant messages, which is at least three times more effective than searching for information by keywords, which is usually done by company employees.