A robot based on a neural network model was developed, which allows for the classification of estimates based on their textual description. Automatic retraining of the model and high-speed operation have made it possible to fully automate this routine process.
How it Works
A file with the estimate is uploaded to the system The neural network performs their classification As a result, it provides the proportions of certain words according to given topics What???s Inside { “TypeOfWork”: “Barrier OZDU-M3 000 ‘CPI’”, “predictions”: [ { “label”: “Installation of low-current systems (including ODS, ASU)“, “probability”: 92.132 }, { “label”: “Installation of ITP”, “probability”: 4.115 }, { “label”: “Installation of heating systems“, “probability”: 0.401 }, { “label”: “Earthworks*”, “probability”: 0.286 }, ] //here we do this and that //it outputs this and that //it outputs this and that Results
Processing time for 100 estimates reduced from 8 hours to 1 minute
The robot’s accuracy is higher than that of a human
Request a Demo Leave your details, and we will send you a demo version of the smart assistant.