Algorithm Development for Implementing the Information Resource Life Cycle Digital Twin

The Concept of an information resource

An information resource (IR) is a collection of data in information systems. For example, a collection of books in a library, customer data in a service center.

IR is used in many different companies engaged in different activities. Someone uses an IR to store a list of products, and someone to store a list of employees.

Being objects of knowledge preservation, IRS serve as data for the construction of new knowledge, which, in turn, can be presented in the form of IR. The use of AI presupposes the possibility of their interpretation either by a person or by computer programs. Information resources exist in electronic form (for example, electronic document files) and other forms.

Each IR is accompanied by a description (specification) containing the data necessary to find it, determine the scope of application.

Since the storage and use of an information resource is expensive, there was a problem of simplifying it. This task has already been achieved by classifying the information resource using the Dublin Metadata Core (DCMI). Due to this, the time and money spent on using the information resource have been reduced.

Dorrer (2009) notes the Dublin Core is a set of metadata elements, the meaning of which is described verbally and fixed in the specifications of the standards defining it. The combination of the values of these elements can be used as a structured description of the content of various kinds of text documents or documents presented in other environments, as well as user requests.

Haytov (2021) consider the product life cycle (product life cycle) is the combination of several processes and phenomena that are repeated at intervals determined by the time of the existence of a particular product, the cycle begins at the moment of conception before its disposal. Such a cycle is a frequent case that is applied to industrial products.

The information resource during its existence goes through the following stages of the life cycle:

• collecting information;
• keeping;
• processing;
• archiving;
• destruction.

Mathematical model of the information resource life cycle

During the life cycle, the relevance of the information contained in the information resource changes. Popov (2011) consider from this point of view, information can be classified as critical, important and unimportant.

Based on the above, the life cycle model of an information resource can be simplified in the form of a Markov chain.

Figure 1 shows a life cycle model of an information resource in the form of a Markov chain.

See Full Size >

There are 5 states in this circuit, which mean:

• S1 - creating an IR;
• S2 - storage and processing of critical importance IR;
• S3 - storage and processing of IR with important information;
• S4 - archiving, storage and processing of IR with unimportant information;
• S5 - removing an IR.

The transition from one state $S_i$ to another $S_j$ in one time step is a random event, the probability of which $p_{ij}$ is determined by the stages of the life cycle of the IR, their "lifetime" and other factors.

The probabilities of the transition of the $i-$th type of resource also determine the average duration of the information resource's stay in each of the selected states.

The $R_{ij}$ resource is characterized by the following parameters:

• the time of receipt of the IR in the information system τ, 0≤τ≤T, where T – is the time of the study of the system;
• the duration of the "life" of the IR TSi - the number of moments of time from the receipt of the resource in the system to the moment of its removal with a probability close to one;
• the volume of the received resource of the i - th type at the time of τ-Vi0(τ), MB;
• vij(t,τ)- the volume of the resource Rij, MB, at the time t≥τ. The values vij(t,τ) form the vectorVi(t,τ)=vi1(t,τ),...,vi5(t,τ), while ∑j=15vij(t,τ)=Vi0(τ),∀t≥τ.
• xij(t-τ)- the probability of finding the resource Rij in the j-degree of relevance at the time t≥τ. Probabilities xij(t-τ) form a vector Xi(t,τ)=xi1(t,τ),...,xi5(t,τ), while ∑j-15xij(t,τ)=1, the initial probability distribution Xi(τ,τ)=1,0,0,0,0.

The resource that has entered the system (in the state $S_1$), is further redistributed between states in proportion to the probabilities of the system being in this state. Based on the above, we will get the following calculation formulas.

The dynamics of the change in the state of the IR is determined by the equations:

$X_i(t+\mathrm{1},\tau )=\left\{\begin{array}{c}\mathrm{0},t\le \tau ,\\ X_i(t,\tau )P_i,t\ge \tau .\end{array}\right.$

$v_{ij}(t,\tau )=\left\{\begin{array}{c}\mathrm{0},t\le \tau ,\\ V_{i\mathrm{0}}\left(\tau \right)x_{ij}(t,\tau ),t\ge \tau .\end{array}\right.$

$i=\mathrm{1},...,n,j=\mathrm{1},...,\mathrm{5}.$

The total amount of resources of the $i-$th type located in the system will be determined by the vector

$V_i^{\Sigma }\left(t\right)={\sum }_{\tau =\mathrm{0}}^T{V}_i(t,\tau ),t=\mathrm{0},...,T_{\mathrm{1}}$,

where $T_1=T+T_{Si};T_{Si}-$ is the lifetime of the resource of the $i-$th type.

The total amount of resources of the $j-$th degree of relevance will be determined by the sum

$u_j^{\Sigma }\left(t\right)={\sum }_{\tau \le t}{\sum }_{i=\mathrm{1}}^n{v}_{ij}(t,\tau ),$

$j=\mathrm{2},...,\mathrm{4,0}\le t\le T$

The calculated volume of the IR of the $j-$th degree of relevance $U_j^{\Sigma }$ for determining the necessary parameters of the storage device is equal to the maximum value of the value $u_j^{\Sigma }\left(t\right)$ at all the time interval under study:

$U_j^{\Sigma }=\mathit{max}{u}_j^{\Sigma }\left(t\right),j=\mathrm{2},...,\mathrm{4,0}\le t\le T$.

Taking this model as a basis, it is possible to build a digital twin of the life cycle of an information resource.

A digital twin is a digital model of any objects, systems, processes or people. It accurately reproduces the form and actions of the original and is synchronized with it.

A digital double is needed to simulate what will happen to the original in certain conditions. This helps, firstly, to save time and money, and secondly, to avoid harm to people and the environment.

The digital double helps to get answers to the "what if" questions, understand the behavior of the system and verify the results of the model. At the same time, different indicators of the real system can be used to verify the model.

Taking the Markov chain model as a basis, the next thing to do is to transfer the model to the AnyLogic simulation environment using the "Process Modeling Library" located in it. First of all, you need to create an imitation of each type of resource according to the Dublin core; a block called Source is suitable for this purpose. This block will generate agents, which, in this case, will represent the type of resources specified by the block. A total of nine resource types will be taken, which are shown in Table 1.

Since different types of resources can come to the IR, with different intensity and at different times, there may be a situation that a certain type of resource may not come to the IR at all, for this, the model will need a side that will temporarily hold the generated agents and a block that will not allow them to exit until this not required. The Queue and Hold blocks will be used for these tasks. The first block will accumulate an unlimited number of agents, while the second block will keep agents in the first block, preventing them from exiting until the Hold block is unlocked, in which case the agents will start exiting the block Queue.