St. Petersburg has increased the speed of search on digital archives tenfold

2023: Improving the efficiency of working with archival documents

The Smart Archive solution, developed by the St. Petersburg Information and Analytical Center, has increased the efficiency of working with archival documents. The search time for information has been reduced several times thanks to the automatic text recognition function . The neural network tool is available for searching on more than 7 million electronic images of documents from the city archive. This was announced by the St. Petersburg Information and Analytical Center on October 30, 2023.

The tasks of the Smart Archive project to introduce artificial intelligence technology into the system for storing and processing archive data include increasing the availability of documents of state archives. As of October 2023, several million archival documents and electronic images are presented on the Archives of St. Petersburg Internet portal. The developers made working with the portal more convenient by reducing the time to search for the necessary information and expanding search capabilities in general.

Created and trained neural networks have expanded the list of portal functions. It added the possibilities of optical recognition of text, extraction of address information and named entities, automatic acquisition of geographical coordinates of addresses with binding to the map. In addition, the service will soon provide the opportunity to automatically form directories, display recommended documents and replenish a single search index based on the extracted information.

With automatic text recognition and content search, users can get the information they need much faster. For example, if it can take 30 minutes to view a document page by page, then the solution allows you to find information 60 times faster - in about 30 seconds.

As a result of the platform update, the general level of satisfaction of citizens with the work of the portal increased - 92% rated the work of the Archives of St. Petersburg as "good" and "excellent." The volume of services provided by the portal has increased - the number of users has grown from 1.8 to 2.5 thousand people per day. At the same time, the load on the archive staff decreased due to the ability of visitors to independently search for data on the text.

In the future, the St. Petersburg Information and Analytical Center plans to add handwriting recognition capabilities to the Smart Archive platform, continue to work on improving the quality of search mechanisms and training neural networks, and introduce a recommendation system based on the behavioral model of users.

Artificial intelligence technologies help optimize document work and open up serious prospects for the archive service. Neural networks will allow us to reduce labor costs for some types of work, while making them faster than a person. Now, with the help of the platform, users will be able to quickly find the necessary documents using various search queries, - said Petr Tishchenko, Chairman of the St. Petersburg Archive Committee.