Notarization with Intelligent Virtual Assistant with Blockchain and NLP technologies (NAVIRO)

Background

Councilbox Technology will research and develop in this project the technologies that will provide a prototype for companies that require corporate meetings, shareholders’ meetings supported by online or face-to-face meetings with electronic evidence and a notarization system, which minimizes the bureaucratic workload and facilitates with recommendations, and all this in an environment of legal security.

The experimental prototype called “Notarization with Intelligent Virtual Assistant with Blockchain and PLN technologies (NAVIRO)” will be implemented based on the development of new systems supported in Blockchain and Natural Language Processing, providing, among others, the following basic functionalities:

Meeting administrators or managers will be able to interact with an intelligent virtual notary through a chat window both by voice and text to obtain information about past meetings and solve certain doubts related to legislation. In addition, the system will improve its behavior through various feedback mechanisms.
The intelligent virtual notary will transcribe through STT techniques the public audio of meetings, generate summaries, and identify the topics discussed automatically. The information generated from the transcripts will be used to expand the knowledge base of the system.
The virtual notary will generate blockchain evidence of the most relevant events of the meetings, such as, for example, the start of a meeting or the issuance of a document and will generate evidence of his interactions with users and of the information he generates himself (transcripts, summaries, etc.).

The implementation of NAVIRO technology (Notarization with Intelligent Virtual Assistant with blockchain and NLP technologies) will open up new markets and possibilities. Technological security, trust and scalability are the pillars and first concerns of major global businesses. Entities such as banks, public administration or the legal sector need to protect data as well as the integrity of their customers.

No company, neither nationally nor internationally, is currently known to offer a notarization system with an intelligent assistant (virtual notary) based on blockchain electronic evidence for online and face-to-face meetings.

Goals

The Notarization system proposed in this project will provide a disruptive solution in the market because it will include a tool to hold corporate meetings (online and/or face-to-face) with legal validity, integrating AI and intelligent notarization.

In order to achieve the specific objectives of the project, the development of the following will be carried out:

Design of the global architecture of the system.
The conversational system (Virtual Notary) allows interaction through text and voice, providing answers on static knowledge bases (limited knowledge on a subset of the Spanish legislation) and on dynamic knowledge bases (information related to previous meetings).
The conversational system is able to improve its response in future interactions through continuous learning from user feedback.
Interaction with the “virtual notary” can be done through text and voice via a web interface executable in a browser.
The “intelligent virtual notary” is able to transcribe and process the audio of meetings, expanding its knowledge base.
The “virtual notary” generates blockchain evidence of the main events of meetings, audio, video, transcripts, and other derived information, and of the notary’s interactions with users.
Experimental prototype available for proof of concept.

Using the so-called “virtual notary” system based on artificial intelligence and blockchain, users of the final product will be able to access, quickly, easily and with electronic evidence, the resolution of doubts and the recovery of information. The assistant will be able to hold basic conversations, provide relevant information about previous meetings and answer basic questions about a relevant subset of Spanish law. In addition, the virtual notary will be able to be present as a listener at the meeting, generating an automated transcript of the meeting in the background. It will detect the most relevant topics of the meeting and generate a summary of the meeting. In addition, it will detect important events during the meeting such as the start of the meeting, opening of votes, list of connected participants, etc. For such events, the notary will generate blockchain evidence unattended and digitally sign it. Once generated, their existence or origin cannot be questioned even by the trusted third party. It will also generate evidence of its interactions with users and the information it generates automatically (transcripts, summaries, etc.).

All these innovative functionalities are currently not available in any known system and imply a clear technological breakthrough that is not possible to achieve with existing technologies on the market.

Innovation level

The proposed solution is positioned as a tool with global reach, allowing access to new markets with no need to change the main features of the tool.

As a result of this project, the following technological advances will be available:

Global cloud architecture
Audio to audio transcription model
Intelligent conversational system
Advanced notary web interface
Massive and low-cost evidence generation system
Electronic evidence retrieval system for evidence verification
API’s for the interconnection of the different modules.
Pre-processing of transcripts
Automatic summary model.
Word extraction model
Automatic question generation model and
Thematic extraction model.
Experimental prototype that will allow the realization of a proof of concept in an environment

Development guidelines

At the functional level, the intelligent virtual notary is conceived as a development with three differentiated components and their respective functionalities:

Generation and processing of transcriptions:

The artificial intelligence models necessary to transcribe the audio of the meetings in Spanish and to obtain information from them, such as summaries and relevant topics. Additionally, digitally signed blockchain evidence will be generated for all the information generated by this module (transcription, meeting topics and summaries).

System with text and voice interaction:

The module in charge of managing the conversation with the user for the retrieval of relevant information based on the context, interpretation, and response to questions from a knowledge base.

Notarization through the generation of blockchain evidence:

Blockchain evidence generation requires the management of an own blockchain network to generate blockchain evidence in the context of massive and variable volumes of requests. In other general open blockchain networks, where desirable properties of the data stored there (such as immutability and timestamping) are robustly guaranteed, the recording of massive volumes of evidence is not possible due to both technical limitations of the networks and economic constraints (writing to them is expensive). In order to be able to grant properties such as immutability and timestamping to the evidence generated in a proprietary blockchain network, it is necessary to consolidate them regularly in a blockchain network that does offer these guarantees.

Milestones achieved by 2022

The following results have been achieved to date:

Analysis:

A state-of-the-art study on AI and blockchain has been carried out.
The requirements for the overall architecture, the assistive system, the scalability of the overall system, the blockchain notarization system and the transcription system are available.
Preliminary description of the proof of concept has been done.

Design:

The choice of technologies for the construction of the AI models and the assistive system has been carried out.
The transcription generation and processing modules, the assistive system module, the blockchain evidence generation and recovery system have been designed.
The design of the global and scalable architecture is available.

Work is currently underway on the development of the blockchain module, training of AI models and creation of APIs.