Activities, documents and semantics

At the moment, modern information systems that simulate activities and document management systems that legally support activities are scattered across different architectural levels, interacting only along the lines of control and accounting. Electronic document management using electronic signature does not solve the problem of the gap between these two levels, providing only the speed and security of document exchange.



In general terms, the task can be formulated as the inclusion of legally significant documents in the digital model of activity as elements that determine the activity.



When solving a problem, we are dealing with several entities



  1. an activity consisting of a partially ordered sequence of actions,
  2. a document that is a set of facts that legally ensure action,
  3. an actor who has the right and performs actions in accordance with the facts stated in the document,
  4. a digital model of activity that prescribes and fixes the actions of the actor.


As noted, the problem is that there is a technological gap between the legally significant fact, semantically defined in the document, and the action in the digital model, the execution of which must be logically / causally consistent with the fact. To implement the activity, there must be an actor at the break point who, on the one hand, understands the semantics of the document, that is, is able to isolate from the document a fact that legally stipulates the action, and, on the other hand, has the right to perform this action. At the moment, only a person can act as such an actor - a subject of law with the appropriate powers, supported by documents (contracts, certificates, diplomas, etc.).



The problem is not solved by using electronic documents, since the meaning of the text in the document file is not available for the digital model of activity. That is, first of all, the problem of “understanding” the digital activity model of the semantics of the document must be solved. At the same time, it is obvious that we should not talk about direct recognition of the content of a text written in a natural language (say, with the involvement of ML-technologies), since the result of such “understanding” cannot be recognized as unambiguous. It is difficult to consider satisfactory and the solution proposed in the blockchain industry, which consists in the complete translation of documents (contracts) into a programming language, that is, the use of so-called “smart contracts”.



The most appropriate solution to the problem of a technological gap can be considered the use of a single, both for describing documents and for modeling activities, a semantic data markup language. In such an ideology, a fact recorded in a document in the form of a statement in this language should be interpreted in a digital model as a command to perform the action prescribed by this fact. In an ideal implementation of this approach, each document, in fact, should be interpreted as a fragment of an activity model. It is extremely important that the language for the semantic description of activities and documents is universal, that is, independent of the subject area, as well as human-readable outside and before specific digital models of activity. The basis of such a language can be a stack of specifications for representing data on the Semantic Web (RDF / OWL, etc.).



However, in addition to the technological barrier, there is also the problem of trust, that is, confirmation of the legal significance of both the document and the actions of the actor, in the capacity of which in the model of activity, on an equal basis with a person, a software agent (a controller executing commands written in the semantic markup language) can also act. The problem of trust splits into two tasks: (1) confirmation of the legal significance of the document (all its facts) and (2) proof of the non-falsification of the action performed by the actor, according to the document.



The first task is solved by signing by the counterparty actors with their private keys of all atomic facts of the document during its preparation and approval. As a result, the document should be a directed acyclic graph, the nodes of which are statements about its atomic facts (including the hashes of the previous facts), and the edges are logical conditioning relations. On the one hand, this document should be used as a fragment of an activity model, on the other, its content can be “read” (programmatically or directly with the eyes) using normative dictionaries, which are also documents. And of course, the document, or rather any fact in the document, can be checked for invariability and authorship. (It should be noted that the document is not edited, changes and additions are made as graph extensions,canceling previous similar statements, that is, the entire history of changes is saved in the document itself).



The task of recognizing the legal significance of the result of an action performed according to a model that includes a document is reduced to two subtasks: (1) verifying the right of an actor - a human or a software agent - to perform an action signed by his private key, (2) excluding accidental or deliberate falsification of the result actions. The first task is to automatically check the relevance of the signatures of the relevant documents in the system. The problem of the validity of the result at the moment can be solved only by reaching a consensus of many independent performers of the activity model, which are nodes of a peer-to-peer network.



So, the ideological basis of the proposed solution is the idea of ​​the need to combine in one digital model of activity: (1) an algorithm of actions, (2) a document that determines the legal conditionality of these actions, (3) an actor of actions, and (4) the activity itself, which, according to in fact, completely goes digital.



The technological basis of the solution is made up of today's standard technologies:



  1. cryptographic methods of encryption and signing of documents,
  2. key management systems,
  3. peer-to-peer networks with consensus transaction validation,
  4. semantic data markup languages.


The technological problems that need to be solved include the development of a stack of specifications for the semantic representation of data to a semantic language for modeling activity (in fact, we are talking about an introduction to the semantics of time).



See also Semantics and Activity



All Articles