Periodic Reporting for period 2 - Signa2.0 (Signaturit)
Okres sprawozdawczy: 2018-10-01 do 2019-12-31
Signa2.0 is a project that establishes a knowledge base that allows the company to offer solutions related to digital transformation by creating new solutions to market needs. Through the development of technological tools (electronic signature, machine learning, OCR, smartcontracts and blockchain solutions) for contractual automation and within the framework of the Project it is intended to solve the problems related to document and contractual management related to the expenses of time, consumables and human resources in any hiring process.
With the traditional paper way, the contractual processes have a longer duration and require a series of administrative-type jobs that make more complicate to manage within an organization regardless of its size. With the solutions developed it is intended to optimize the entire contracting process, automating and adding a security bonus.
The potential impact that these tools can have for the management of the contractual process and document management makes the solution establish the appropriate basis in the contractual field to optimize the creation, management and execution processes. With these solutions it is intended to achieve an automation of the execution of the contract and the registration of the related data to allow the storage of the internal procedures of the user and, through data analysis, obtain greater knowledge of their contracts.
Within the framework of the implementation of solutions related to the trust services of the eIDAS Regulation, blockchain solutions, smartcontracts, Machine Learning and OCR it can be concluded that the benefits can be apart from tangible cost savings, avoiding shipments, or reducing the consumption of ink or paper:
- Avoid displacements and save time in the signature processes
- The digital archive of signed documents.
- An optimization of the document management process thanks to storage in digital format.
- Offer greater security in the signature processes since there is more evidence throughout the signing process.
- Optimization and automation of contract execution processes and, therefore, those related to customer service.
- Greater reliability and reduction of risks in the execution and interpretation of contracts.
- Increased knowledge of the needs and habits in the signature processes of the clients and prediction of needs and trends.
- Classification of documents according to their typology.
The Signa 2.0 main objectives are provide QES, official timestamp and eDelivery services, the development of a commercial Smart Contract, development of an Optical Character Recognition of European Ids and development of Machine Learning within a single contract and document management platform that reaches the entire contraction process of a company.
With the traditional paper way, the contractual processes have a longer duration and require a series of administrative-type jobs that make more complicate to manage within an organization regardless of its size. With the solutions developed it is intended to optimize the entire contracting process, automating and adding a security bonus.
The potential impact that these tools can have for the management of the contractual process and document management makes the solution establish the appropriate basis in the contractual field to optimize the creation, management and execution processes. With these solutions it is intended to achieve an automation of the execution of the contract and the registration of the related data to allow the storage of the internal procedures of the user and, through data analysis, obtain greater knowledge of their contracts.
Within the framework of the implementation of solutions related to the trust services of the eIDAS Regulation, blockchain solutions, smartcontracts, Machine Learning and OCR it can be concluded that the benefits can be apart from tangible cost savings, avoiding shipments, or reducing the consumption of ink or paper:
- Avoid displacements and save time in the signature processes
- The digital archive of signed documents.
- An optimization of the document management process thanks to storage in digital format.
- Offer greater security in the signature processes since there is more evidence throughout the signing process.
- Optimization and automation of contract execution processes and, therefore, those related to customer service.
- Greater reliability and reduction of risks in the execution and interpretation of contracts.
- Increased knowledge of the needs and habits in the signature processes of the clients and prediction of needs and trends.
- Classification of documents according to their typology.
The Signa 2.0 main objectives are provide QES, official timestamp and eDelivery services, the development of a commercial Smart Contract, development of an Optical Character Recognition of European Ids and development of Machine Learning within a single contract and document management platform that reaches the entire contraction process of a company.
The work performed from the beginning of the project to the end of the period can be summarized into four main fields: Trusted services, OCR Engine, machine learning and Smart Contract solution. Signaturit has passed the relevant audits to be able to offer the services of qualified electronic signature, qualified electronic delivery and qualified time stamping. We have managed to be on the European lists of trusted service providers for these services.
Signaturit’s OCR Engine is the first beta version for automatic recognition and extraction of data from ID documents. Currently the implemented OCR engine is able to identify ID documents within an image and Detect and read the contained text instances.
The initial goal was two-fold, first of all we wanted to know whether a document would be signed or not and if it was going to be signed we wanted to predict the “when”. However, after analysing Signautirt’s platform data we discovered that when the documents get signed, it happens during the first day. It was concluded that trying to predict the day a document would be signed, given it was happening almost always during the first day, was an irrelevant task.
Therefore, a slight deviation from the initial plan was performed: a side-project that automatically labels documents designed in collaboration with expert knowledge. The results of this Project have been fruitful in two directions. The Automatic Labeler of documents produces internal insights about the platform’s usage, and secondly it is also an appealing feature for clients that want to categorize and organize the documents they send through Signaturit’s platform.
In terms of blockchain, we studied the regulation of personal data to create a system capable of linking any Signaturit user action with transactions and evidences on blockchain. This establish the fundamentals for a blockchain based identification system of our users. We finally build a system capable to manage a rate of transactions per second (TPS) to be stored on blockchain which is very high compared to public blockchains.
We secured all of our core events with an immutable timestamp and recoverable evidence directly on blockchain. Being able to not just provide a proof but also the entire history of an user gives us the possibility to build even more products. We built a set of smart contracts representing all the Signaturit platform entities like signatures, documents, etc... and we also decided to make them open source following in deep the blockchain philosophy of being transparent and public auditable.
Signaturit’s OCR Engine is the first beta version for automatic recognition and extraction of data from ID documents. Currently the implemented OCR engine is able to identify ID documents within an image and Detect and read the contained text instances.
The initial goal was two-fold, first of all we wanted to know whether a document would be signed or not and if it was going to be signed we wanted to predict the “when”. However, after analysing Signautirt’s platform data we discovered that when the documents get signed, it happens during the first day. It was concluded that trying to predict the day a document would be signed, given it was happening almost always during the first day, was an irrelevant task.
Therefore, a slight deviation from the initial plan was performed: a side-project that automatically labels documents designed in collaboration with expert knowledge. The results of this Project have been fruitful in two directions. The Automatic Labeler of documents produces internal insights about the platform’s usage, and secondly it is also an appealing feature for clients that want to categorize and organize the documents they send through Signaturit’s platform.
In terms of blockchain, we studied the regulation of personal data to create a system capable of linking any Signaturit user action with transactions and evidences on blockchain. This establish the fundamentals for a blockchain based identification system of our users. We finally build a system capable to manage a rate of transactions per second (TPS) to be stored on blockchain which is very high compared to public blockchains.
We secured all of our core events with an immutable timestamp and recoverable evidence directly on blockchain. Being able to not just provide a proof but also the entire history of an user gives us the possibility to build even more products. We built a set of smart contracts representing all the Signaturit platform entities like signatures, documents, etc... and we also decided to make them open source following in deep the blockchain philosophy of being transparent and public auditable.
Regarding the provision of trust services, there is no major innovation because these are services that are regulated by European regulation and by ETSI EN regulation, so in order to offer these services it is necessary to adapt to the provisions of the regulations.
The key improvement has been to increase the accuracy of ID readability beyond the commercial alternatives. Usually modern OCR engines that go beyond reading text under very controlled conditions lack of performance and accuracy because of their broad scope. By narrowing the context to ID documents, Signaturit has achieved better results for that specific use case. A robust ID verification system that is not sensible to external conditions is a product that is not only appealing for Signaturit’s use case, but also to a broader set of ID verification use cases. As a matter of fact, communications with other entities that are focused on ID verification have been initiated so Signaturit can be part of the loop in defining the desired standards of such systems.
During the past recent years, the Legaltech sector has been adopting more and more solutions that come from Machine Learning and its power to automatize what would represent arduous and manual analysis. Having built a system that’s able to predict to some extent the correctness of a legal document Signaturit has gone beyond the current market solutions. With the aids of a Machine Learning system such as the one implemented during this project; the workload of the legal figure can be minimized. This would have the potential to have an impact on the economic aspect of such services.
We strongly believe being experts in blockchain technology now. We built a system of transactions management that was not existing before in our knowledge and which is fast, scalable and reliable. Also our transactions management system is able to retry transactions in the case of some errors (like network congestion) and able to provide complete traceability of transactions.
The key improvement has been to increase the accuracy of ID readability beyond the commercial alternatives. Usually modern OCR engines that go beyond reading text under very controlled conditions lack of performance and accuracy because of their broad scope. By narrowing the context to ID documents, Signaturit has achieved better results for that specific use case. A robust ID verification system that is not sensible to external conditions is a product that is not only appealing for Signaturit’s use case, but also to a broader set of ID verification use cases. As a matter of fact, communications with other entities that are focused on ID verification have been initiated so Signaturit can be part of the loop in defining the desired standards of such systems.
During the past recent years, the Legaltech sector has been adopting more and more solutions that come from Machine Learning and its power to automatize what would represent arduous and manual analysis. Having built a system that’s able to predict to some extent the correctness of a legal document Signaturit has gone beyond the current market solutions. With the aids of a Machine Learning system such as the one implemented during this project; the workload of the legal figure can be minimized. This would have the potential to have an impact on the economic aspect of such services.
We strongly believe being experts in blockchain technology now. We built a system of transactions management that was not existing before in our knowledge and which is fast, scalable and reliable. Also our transactions management system is able to retry transactions in the case of some errors (like network congestion) and able to provide complete traceability of transactions.