At the invitation of the KTU Artificial Intelligence Centre, university researchers together with “UAB Topo Grupe” implemented a project during which a commodity information collection system was developed.
The goal of the project implemented last year was to create an intelligent system capable of autonomously collecting and recognising the necessary information (textual and visual) about products from various data sources. An artificial intelligence algorithm should be integrated into the system to process and combine the collected information from multiple sources, predict missing information, provide the required information to the end-user in a structured way and transfer it to other external systems.
According to scientists, data about commodities is collected and analysed with the help of software robots. The parameters are grouped and similar objects are recognized by applying the sentence transformation (sbert.net) model, comparing each parameter with ICECAT parameters and selecting parameters if the similarity match exceeds the 90 per cent threshold.
The system also uses artificial intelligence algorithms that allow predicting missing information, providing the necessary information to end users in a structured manner and transferring it to other external systems.
The task presented by the “UAB Topo grupe” was implemented by the team of KTU Faculty of Informatics: lecturer Donatas Sandonavičius, associate professor Andrius Krisčiūnas, an alumnus of program systems engineering Vilius Kerutis and graduate of information systems engineering program Gediminas Šmita.
The value of the project “Autonomous Product Information Collection System” (project no. PP2022/55) was EUR 50,000.