Neural network for the document circulation system

AI & Machine Learning — incorrectly specified categories?

Project translated automatically. Log in or register, to view the original Hello to you!We are developers of the document circulation system from Lithuania.We want to add it to AI who would know: 1 .Determine the type of document (which for the document: Invoice, Contract, Request, Request for a vacation, etc.To establish a number of requests: from whom he is (who is the counterpart, who is our employee), the number of the document, the date of the document.2ndCheck out some simple errors in the documents.For example, the contract does not correctly specify the telephone number.Documents that type to be determined can be both in parsered formats (Word or readable PDF), and in the form of a picture (from a scanner, for example).That is, as I understand, it needs some OCR solution.It can be a ready solution if you can buy it, rather than pay a rental for reading each document.We don’t have any machine training specialists.We don’t really understand this.Therefore, there is an idea that it is more reasonable to hire someone from the side than to understand themselves.We write the system on C#, we work with .NET Core.It is better for us to have something close to it.But not principally.There may be a module that gives us the result in the form of a response (i.e., it reports the type of document and from whom it is received).In the proposal, please write in general features: using what tools (framework, languages or ready-trained networks) you would solve such a task and what architecture would be.What about money and deadlines?

Proposals 3