Constructing a Quantitative System of Intelligent Trial Standards for Civil and Commercial Law Cases Based on Network Flow Algorithm in Intelligent Judicial Environment

With the rapid development of information technology and the deepening of digital transformation, the judicial field is facing unprecedented challenges and demands. In the current context, the judicial system not only needs to cope with the increasing number of cases and processing pressure, but also needs to improve its efficiency, transparency and intelligence while ensuring judicial justice [1-4]. Social demand for judicial services is also increasingly diversified and personalized, requiring the judicial system to provide more accurate and efficient legal services. With the continuous progress of technology, the application of artificial intelligence in various fields has become more and more extensive, and judicial trial is no exception [5-8].

In the past, determining a case often required judgment based on human subjective consciousness, but with the development of AI technology, it is becoming increasingly capable of helping us more accurately determine the fairness and legality of a case [9-11]. Through the use of artificial intelligence, it may change the traditional way of trial, reduce the work pressure of the trial judge, improve the work efficiency, and at the same time the use of artificial intelligence, based on the analysis and comparison of the case issues, to make an accurate and fair judgment results, which can correct the trial of the appearance of the wrongful decision or unfair judgment [12-15]. Therefore, the introduction of artificial intelligence into the court trial system, is the general trend, but also the direction of the court to trial-centered reform and development [16-17]. And at the same time, we also need to clearly recognize the risks and limitations of artificial intelligence, and carry out the necessary supervision and error correction [18-20]. In this context, it is particularly important and urgent to construct a quantitative system of judicial trial standards that integrates artificial intelligence (AI) technology.

In this paper, we deeply analyze the difficulties in the application of AI in judicial trials, and in order to improve the efficiency of civil and commercial law case trials, we design an intelligent trial standard quantification system based on the network flow algorithm. The case trial is regarded as a minimum cost flow problem, and the network flow function is assigned a real number to measure whether it satisfies the flow conservation and capacity constraints, so as to calculate the total cost of the flow. A feasible pairwise solution is obtained by computing the optimal feasible solution of one basis of the problem and subjecting it to complementary relaxation conditions. A generalized correlation matrix is also introduced to seek out base arcs or base relaxation variables to efficiently handle the case trial problem. Based on this algorithm, a quantitative system of intelligent case trial standards integrating case filing, file generation, evidence presentation and cross-examination is constructed, and the trial quality and efficiency of civil and commercial law cases under the system are evaluated through practical applications.

2

Shortcomings of Artificial Intelligence in the Administration of Justice

Despite the fact that AI technology has now achieved relative maturity in its application, there are still some application challenges in the judicial environment, which will be briefly described in this section from the following aspects.

2.1

Low Signal-to-noise Ratio and Data Sparsity

Judicial data is low in quality and low in quantity, and the current body of judicial data is insufficient. Although the China Judicial Documents Network (CJDN) is the largest judicial database in China and provides the main legal data resource for the AI case handling system. However, according to relevant statistics, the documents made public on the China Judgment and Decision Network are about half of all judicial decisions in China, and about half of the other information is not made public on the system. In addition to this, due to the uneven economic development of the country’s regions, the level of smart court construction also varies. Such differences are reflected not only in the number and proportion of judicial documents made public, but also in the completeness and depth of judicial data.

2.2

The Infeasibility of Dialectical Reasoning Systems

In judicial trial practice, AI mainly simulates the judicial reasoning process to perform logical reasoning tasks. However, like many AI applications, the current judicial AI is still at the stage of weak AI, Its functions are limited to simple tasks predefined by the developer and can only perform basic formal reasoning. In the design of the AI system, the setting of the logical structure can only be limited to the scope of formal reasoning, which cannot cover the structure of judicial dialectical reasoning. As for more complex judicial dialectical reasoning, the current stage of judicial artificial intelligence is not yet capable of doing so.

2.3

Weakness of Core Trial Control

With the improvement of technology, the application scenarios of artificial intelligence in judicial trials are increasing. This greatly reduces the burden on judges, but also poses the hidden danger that judges may become dependent on artificial intelligence systems. Letting artificial intelligence participate in the judge’s trial work, or even replace the judge to make a judgment, which seriously violates the basic principle of independent trial, not only may make the judge’s trial responsibility difficult to pursue, but also will have an impact on the judge’s dominant position in the trial, which in turn affects the justice and efficiency of the judiciary.

2.4

Fragmentation of Legal Development

Changes in the law often lag behind, making it difficult to solve the problems caused by the application of artificial intelligence in a timely manner. At the same time, the rapid development of artificial intelligence is also the lack of timely legal regulation guidance, which leads to the application of artificial intelligence technology in the judicial trial of instability and insecurity. Theory without practice as the foundation, it seems empty. And if the practice lacks the guidance of theory, it will inevitably appear blind and spontaneous. In the judicial trial, the addition of artificial intelligence is both an opportunity and a challenge. The two complement each other, but also each other constraints.

2.5

Insufficient Supply of Law-intelligent and Law-integrated Personnel

In judicial practice, the judicial intelligence systems or platforms used by the courts are not independently developed, but are accomplished through technology outsourcing and the purchase of third-party services, etc. Within the judicial system, there is still a shortage of composite talents who know both law and technology.

3

Distribution Network Simplex Algorithm

Network flow theory mainly discusses two types of problems maximum flow problem [21] as well as minimum cost flow problem [22], in which minimum cost flow problem is a core problem in network optimization. The traditional network flow theory can solve many practical problems, but it also has its limitations, for some special problems can not be described much less solved. In this paper, for the above intelligent judicial trial problem, we propose the allocation network flow algorithm, which transforms the case elements into network nodes, and designs the reasonable weight and capacity parameters to realize the quantitative analysis of the facts and legal relations of civil and commercial law cases.

3.1

Minimum Cost Flow Network

Definition 1: A general minimum cost flow network [23] is defined as follows:

Define the following weight functions on a directed graph G = (V, A) with V as the set of nodes and A as the set of arcs: 1)

C : A → R is the weight function on the arcs, and the weight C(i, j) corresponding to arc (i, j) ∈ A is denoted as c_ij and is called the cost or fee per unit flow for arc (i, j).

2)

L : A → R is the weight function on the arcs, and the weight L(i, j) corresponding to arc (i, j) ∈ A is denoted as l_ij and is called the lower bound on the capacity of arc (i, j).

3)

U : A → R is the weight function on the arcs, and the weight U(i, j) corresponding to arc (i, j) ∈ A is denoted as u_ij and is called the upper bound on the capacity of arc (i, j).

4)

D : V → R is the weight function on the vertices, and the corresponding weight D(i) of node i ∈ V is denoted as d_i, which is called the supply and demand at vertex i.

The network formed at this point is called a capacity-one-cost network, or it can be directly referred to a flow network or a network.

For a capacity-one-cost network $N = (V, A, C, L, U, D)$ , a flow x on it refers to a function from the set of arcs A of N to R, i.e., for each arc (i, j) ∈ A, a real number x_ij is assigned to be called the flow of the arc (i, j), if the flow x satisfies the following flow conservation condition: (1) $\sum_{j (i, j) \in A} x_{i j} - \sum_{j (j, i) \in A} x_{j i} = d_{i} \forall i \in V$

and satisfy the capacity constraint: (2) $l_{i j} \leq x_{i j} \leq u_{i j} \forall (i, j) \in A$

Then x is said to be a feasible flow, and x is said to be a pseudo-flow if x satisfies only (2) and not necessarily (1).

In general, it is always possible to study the network of L ≠ 0 as a network of L = 0, so assuming $L = 0 (l_{i j} = 0)$ and shorthanding the capacity-one-cost network of L = 0 as $N = (V, A, C, U, D)$ , (2) can again be written as: (3) $0 \leq x_{i j} \leq u_{i j} \forall (i, j) \in A$

The total cost of stream x is defined as: (4) $C (x) = \sum_{(i, j) \in A} c_{i j} x_{i j}$

3.2

Distribution Network Simplex Algorithm Construction

This section focuses on the minimum cost flow problem for a distribution network containing one source point and containing multiple equilibrium points, multiple distillation points, and multiple end points. The general form of the distribution network is shown in Figure 1.

The mathematical model of the minimum cost flow problem for the distribution network flow model is as follows: (5) $\begin{array}{l} \min c_{s} x_{s} + \sum_{(i, j) \in A} c_{i j} x_{i j} \\ s . t . x \in F_{D} \\ x_{j} \geq d_{j} \forall j \in V_{T} \\ x_{s} \leq u_{s} \end{array}$

where F_D is the set of all feasible flows to be satisfied, d_j denotes the demand at the jrd endpoint, u_s denotes the finiteness of the source material, c_s denotes the unit cost of the source material, and c_ij denotes the unit cost of the flow passing through arc (i, j).

To address the above problems, in this paper, we design special simplex methods by combining the network characteristics of the distribution network optimization problem, and we refer to such algorithms as distribution network simplex algorithms [24]. The general idea of simplex algorithms for linear programming problems is to first find a base feasible solution to the problem and end it if it is the optimal solution, otherwise choose an incoming base variable and an outgoing base variable. Improve to another base feasible solution by rotational transformation. This is iterated until the number of tests are all less than or equal to zero. At this point the optimal solution is obtained, i.e., the simplex algorithm always maintains the original feasibility of the solution, but may not be pairwise feasible. When the complementary relaxation condition holds, once the optimal solution is reached, a feasible pairwise solution is obtained at the same time.

3.3

Generalized Correlation Matrices and Base Feasible Graphs

Definition 2: For the minimally allocated cost flow problem (5) defined on $G = (V, A)$ , the subgraph of G corresponding to x is said to be the base feasible graph of (5), assuming that x is a base feasible solution.

Structure of the base feasible graph:

For convenience in dealing with the problem, assume that u_s = + ∞, u_ij = + ∞, holds for ∀(i, j) ∈ A. Since problem (5) contains D points and holds for ∀i ∈ V_D, such that $k_{i} = \sum_{j \in L (i)} k_{i j}$ , then k_i may or may not be 1. When k_i = 1, the value of the stream flowing into point i is equal to the value of the stream flowing out of point i, the problem can be described by a general correlation matrix. It is this special case that is dealt with in When k_i is not 1, then the general correlation matrix can no longer be used to describe the problem, and in order to solve this case, the generalized correlation matrix is introduced [25].

Definition 3: The generalized correlation matrix M, the generalized correlation matrix M of Fig. G = (V, A) is defined as follows, and M is a n × m matrix, i.e., $M = {(m_{i i})}_{n \times m}$ , where: (6) $m_{i t} = {\begin{matrix} - 1 & \exists j \in V, l = (i, j) \in A \\ 1 & F o r i \notin V_{D}, \exists j \in V, l = (j, i) \in A \\ k_{i} & F o r i \in V_{D}, \exists j \in V, l = (j, i) \in A \\ 0 & O t h e r \end{matrix}$

n is the number of points in the network, m is the number of arcs in the network, and for ∀i ∈ V_D, there is $k_{i} = \sum_{i \in L (i)} k_{i j}$ .

For each j ∈ V_T, introducing the slack variable t_j, $\sum_{i \in E (j)} x_{i j} \geq d_{j}$ becomes: $\sum_{i \in E (j)} x_{i j} - t_{j} = d_{j}$ , where t_j ≥ 0, thus becomes: (7) $\begin{array}{l} \min c^{T} x \\ s . t . M x - Q t = d \\ D x = 0 \\ x \geq 0, t \geq 0 \end{array}$

Here x is the flow vector including x_s, c is the cost vector, M is the generalized correlation matrix, and t is the slack variable corresponding to the points in V_T. It is usually assumed that c_s > 0, c_ij > 0. For ∀i ∉ V_T there are d_i = 0, $Q = (\begin{matrix} 0 \\ I \end{matrix})$ , I corresponding to V_T, 0 corresponding to V\V_T, and Dx = 0 corresponding to V_D. For each i ∈ V_D, since there is $k_{i} = \sum_{j \in L (i)} k_{i j}$ , it is easy to see that for each i ∈ V_D, $\bar{M} = (\begin{matrix} M & - Q \\ D & 0 \end{matrix})$ is row-full rank and $M = {(m_{i j})}_{n \times m}$ , $Q = {(q_{i j})}_{n \times p}$ , $D = {(d_{i j})}_{q \times m}$ , $\bar{M} = {({\bar{m}}_{i j})}_{(n + q) \times (m + p)}$ , where n is the number of points in G, m is the number of arcs (including x_s), p is the number of points in V_T, and q is the difference between the number of outgoing arcs in V_D and the number of points in D.

Definition 4: Active and Idle Points, assuming x is a base feasible solution of (7), a point i is said to be idle if there is a point i through which no positive flow passes, otherwise it is said to be active.

Definition 5: Branches of G_B and Extended Branches of G_B: By removing arc x_s and point s from G_B, G_B is decomposed into parts, each of which is connected, and each of which is called a branch of G_B, and by adding arc x_s and point s to each branch, it becomes an extended branch of G_B.

3.4

Network Simplex Algorithm for Distributing Network Flows

Assuming the non-degenerate case, G_B is the base feasible graph corresponding to the given base feasible solutions x and t. The network simplex algorithm for the assignment network flow problem is as follows:

Step 1: Compute the pairwise basis solutions y and u.

Step 2: Substitute the feasibility conditions using y and u obtained in the first step, if both are satisfied, the optimal solution is obtained and the computation is finished, otherwise a non-base arc (i, j) or a non-base relaxation variable t_i that does not satisfy the feasibility conditions is chosen as the incoming base arc.

Step 3: Find an out-of-base arc or out-of-base relaxation variable by rotational transformation, update x and t, turn to the first step and proceed to the next iteration.

Notice that the two points of the incoming base arc can only be in the same branch or both branches, so the outgoing base arc must also consider only this branch or both branches, and other branches need not be considered.

4

Quantitative System for Intelligent Trial Standards Based on Network Flow Algorithms

Based on the distribution network simplex algorithm described above, this section constructs an intelligent trial standardization system covering case acceptance, file generation, evidence analysis and fact finding, trial management, and recommendation of adjudication results, and carries out piloting and application to assess the effectiveness and reliability of the system.

4.1

Intelligent Case Initiation

Upon receipt of a litigant’s application for filing a case, the filing system of the Intelligent Litigation Platform will automatically recognize the litigant’s request, review the request based on the case information included, and if it meets the statutory filing criteria, the Intelligent Filing System will make a decision to file the case, and if it does not meet the criteria, the case will not be filed. In addition, the system can also intelligently review whether the format of the litigant’s application documents for filing meets the writing requirements, to avoid the occurrence of cases that could have been filed on the spot but could not be filed due to errors in the format text or irregularities in writing, making the filing system services not only intelligent. The application of artificial intelligence in the filing stage makes the waiting time for the parties to file a case greatly shortened, and also greatly improves the efficiency of the court’s case filing work, helping greatly.

4.2

Intelligent File Generation

The electronic file is the foundation of the informationization and intelligent construction of the judicial system, and the people’s courts at all levels have popularized and made in-depth use of the electronic file through publicity. When the petition is submitted by the parties in the filing system, the case officer can immediately scan the relevant materials to generate an electronic version, and through the two-dimensional code is associated, so that the relevant filing information can be automatically backfilled in the system without manual input. Compared with the previous manual typing entry, in the archives only when the file is scanned and other time-consuming and labor-intensive practices, the current electronic file of intelligent synchronization with the case generation can be said to be a lot easier.

4.3

Intelligent Proof-examination

In this paper, the electronic evidence system in the intelligent trial standard quantitative system has completely revolutionized the way of evidence presentation and questioning in the traditional court trial. The parties sitting in the seat click on the computer monitor or voice wake-up system to give instructions can be accessed to need to show such as text, pictures, audio and video and other electronic evidence, in the bench and the other party’s display can be synchronized display, unified playback, significantly improve the efficiency of the trial, to maintain the order of the trial. Moreover, when the evidence is displayed, both parties can circle the mark on the display screen, the system will automatically draw the attention of the judge and the other party, so that targeted comments can be made. At the same time, in order to ensure safety, the judge’s marking content will not be displayed to the parties to the trial, this function has a strong humane and practical considerations.

4.4

Intelligent Trial Management

In the process of building an intelligent trial standard quantitative system, this study not only applies the network flow algorithm to each node of the trial process, the trial management business module is also embedded into the intelligent work case handling platform within the court. The work standards, trial requirements, time limits and other management matters involving each node are incorporated into the unified intelligent management. With the addition of algorithms and big data technology, the court can then determine whether the workload of judges has been saturated through statistical system data. At the same time, based on the judicial big data to distinguish the degree of difficulty of the case, simple cases and difficult cases will be scientifically and reasonably matched, and then assigned to the judges. This ensures the quality and efficiency of trials and breaks the pattern of traditional court administration.

4.5

Intelligent Sentencing Recommendations

Distribution network flow algorithms combined with artificial intelligence technology can intelligently select and analyze the circumstances of a case regarding sentencing and can make intelligent predictions about the length of a sentence. Future sentencing activities, especially sentencing in plea bargaining cases, can improve the accuracy of sentencing recommendations with the help of artificial intelligence trial systems. At present, further accelerating the research and development and popularization of intelligent assistive systems for sentencing is a sure way to enhance the objectivity, precision and credibility of sentencing recommendations.

5

Application of a Quantitative System for Intelligent Trial Standards Based on Network Flow Algorithms

In this paper, the network flow algorithm described above is analyzed for link utilization and data caching performance of the network. In order to highlight the advantages of the algorithms in this paper, they are compared with existing network flow algorithms. Specifically, the two-commodity maximal flow algorithm (S-TMF) is proposed for static networks. The two-source flow solving method (S-FIFO) that combines the commodity prioritization policy with the static network widening path algorithm. A two-commodity flow solution method (D-FIFO) that combines the commodity prioritization policy with the time-varying network single-source maximum flow algorithm. All simulation tests are performed on HP workstation Z640 (Intel Xeon E5-2630 v3 2.40GHz, 16GB RAM, O.S. Windows 7 Professional64bits).

The link utilization of each algorithmic network is obtained as shown in Figure 2. It can be seen that the curves for all methods increase rapidly until 90 minutes and then stabilize over time. Meanwhile, the algorithms in this paper can obtain higher link utilization compared to S-FIFO, S-TMF and D-FIFO methods, reaching 84.39% at 220 minutes. In addition, in the comparison algorithm, the D-FIFO method can obtain higher link utilization compared to S-TMF and S-FIFO, which further indicates that the time-varying graph can fully utilize the link resources of the network with the storage resources over time.

In order to further analyze the consumption of storage resources by the algorithm proposed in this paper, this subsection compares the method of this paper with the D-FIFO method in terms of maximum node cache and average node cache. Fig. 3 shows the variation of maximum node cache and average node cache with the increase of a given time range. The results show that the maximum amount of data stored by a node is closely related to the size of the given time range. This is because network links are usually intermittently connected and a long network topology will result in more data being stored in the system. As a result, the network nodes will consume more storage resources. Compared to the D-FIFO method, the algorithm in this paper needs to consume slightly less storage resources, which is 1.66G lower than it at 220 min. this is because the algorithm in this paper requires less memory and consumes less storage resources at the cost of obtaining a relatively larger network stream. It shows that the algorithm in this paper is able to operate with limited memory resources and is suitable for dealing with large-scale network flow problems. In terms of the average node cache, the average node cache gradually increases with the increase of the given time range, which further confirms the effect of the given time range on the node storage resource occupation.

5.1

Evaluation of the Impact of Smart Trials on Judicial Efficiency

The constructed quantitative system of intelligent trial standards for civil and commercial law cases allows searching by cases key words to extract the basic situation of civil and commercial law cases, including the case intake, the total amount of filed subject matter, and the closure of cases. As people’s legal awareness and legal thinking continue to increase, the average number of cases received by the court is on a rapid upward trend. In front of such a huge number of cases, it is often difficult for judges to strike a balance between fairness and efficiency when handling cases. The intelligent trial standard quantification system of this paper was formally applied in Court A at the end of 2018, and this section uses the above model to statistically analyze the basic situation of civil and commercial law cases in Court A in 2019~2024, in order to analyze the impact of the intelligent trial standard quantification system of this paper on the judicial efficiency of this court after it is put into use.

Figure 4 shows the basic situation of civil and commercial law cases and changes in judicial efficiency in Court A from 2019 to 2024. As can be seen from the figure, the number of cases received and the number of cases closed in Court A increased continuously from 2019 to 2023, and it did not drop a little until 2024. In contrast, the judicial efficiency of Court A, in both years 2019~2024, shows an increasing trend, from 66.31% to 84.43%. This shows that the judicial intelligent trial system constructed on the basis of the above model greatly alleviates the contradiction of “too many cases, too few people”, and can help judges deal with cases, thus improving the efficiency of the trial and saving the resources of the trial.

5.2

Study on Optimizing the Effectiveness of Trial Management

The traditional trial management in Court A, such as the deviation between the perspective of case management and the perspective of the cause of the case valued by the commercial subject, the deviation between the case handling thinking and dispute resolution thinking, and the existence of management blind zones in the flow of transactional processes in the case handling process, all lead to perceptual errors of the supply side and the demand side of the judicial product. This section selects the top 20 types of cases that account for the number of cases in Court A in 2024, and specifically analyzes the average trial time of each type of case, in order to evaluate the effectiveness of the intelligent trial standard quantitative system constructed in this paper in the trial management of civil and commercial law cases.

The number of civil and commercial law cases and the average trial days statistics are shown in Figure 5. After the implementation of this paper’s intelligent trial standard quantitative system, Court A vigorously promotes the diversion of cases into simple and complicated cases, and the simple cases are quickly tried and complicated cases are carefully tried, so that most of the simple cases can be quickly finished. According to the full-sample assessment method and the average value measurement method of quality and efficiency assessment, the cases that account for a higher percentage of cases are particularly important, and the lower their average trial time, the greater the momentum to drive down the overall average trial time. Such as the number of cases accounted for the top three of the three types of cases average trial time were 28.86, 24.89, 20.36 days. Compared with the implementation of this paper before the implementation of intelligent trial standard quantitative system, the average trial time of the three types of cases decreased by a larger margin, indicating that this paper intelligent trial standard quantitative system for the judicial trial management of Court A showed better auxiliary effect, greatly reducing the case trial time.

5.3

Pre-establishment of Dispute Resolution Methods

In order to systematically study the effect of the intelligent trial standard quantification system under the algorithm of this paper on the dispute processing of civil and commercial cases, this section takes the financial loan contract dispute case as an example, and utilizes the method of this paper to make a prediction of the dispute resolution method.

Financial loan contract dispute case there is a characteristic: that is, as long as the defendant to the case of the general dispute between the two sides is not big, and most of them can mediate. So for this kind of case can be analyzed by the nature of the party’s work, income, psychological expectations, the amount owed, interest calculation, repayment plan, family situation, children’s situation and litigation, recorded as dispute resolution 1 ~ 9. Calculate which dispute resolution is more suitable for the handling of this case, to achieve the highest efficiency of the trial and the implementation of the effect.

Regarding the measurement of dispute resolution, assuming a peak value of 100, we analyze the factors of different case closure methods based on 100 class cases, as well as the mediation closure impact factor peak size to determine the mediation program, due to the limited number of statistics, take the valuation for measurement. The results of the measurement of dispute resolution methods are shown in Figure 6. As can be seen from the figure, this paper’s intelligent trial system for financial loan contract dispute cases in different dispute resolution measurements between 20 ~ 70, of which “repayment plan” dispute resolution measurements for the largest value of 65, that is, indicating that in this financial loan contract dispute case to “repayment plan That is to say, in this financial loan contract dispute case with “repayment plan” as the dispute resolution method can maximize the efficiency of trial and implementation. Adopting the dispute resolution method predicted by the intelligent trial standard quantitative system in this paper, this case has achieved very good regulation effect.

6

Conclusion

The article learns that there are problems such as low signal-to-noise ratio of judicial data and the sparseness of artificial intelligence in judicial trial by reviewing relevant literature. For this reason, it constructs a quantitative system of intelligent adjudication standards for civil and commercial law cases based on network flow algorithm. Firstly, the case elements are transformed into network nodes, and the total cost of the network flow is measured through the minimum cost flow network. Then combining with the simplex algorithm, a base feasible dyadic solution of the problem is sought by utilizing the rotation change. Finally, the generalized association matrix is introduced to represent the connection relationship between the network nodes and edges to consolidate the feasibility of the algorithm. The constructed intelligent trial standard quantization system can collaborate the efficient judicial trial process from the aspects of case filing and file generation. The system of this paper is applied to Court A, and the following results are obtained: the network link utilization rate of this paper’s algorithm is 84.39% at 220 minutes, which is much higher than that of the comparison algorithm, and it can deal with large-scale network flow problems by consuming less storage resources. The judicial efficiency of Court A increases from 66.31% to 84.43% with the intervention of the intelligent trial standard quantification system. The trial time of cases is significantly shortened, and the average trial time of the top three types of cases is 28.68, 24.89, and 20.36 days, respectively. For financial loan contract disputes, this paper system calculates that the “repayment plan” dispute resolution method has the largest factor value, which is the best dispute resolution method for this case. The network flow algorithm adopted in this paper can provide a new technical path for the construction of intelligent justice.

Idioma:: Inglés

Calendario de la edición:: 1 veces al año
Temas de la revista:: Ciencias de la vida, Ciencias de la vida, otros, Matemáticas, Matemáticas aplicadas, Matemáticas generales, Física, Física, otros

RSS Feed de revista

Constructing a Quantitative System of Intelligent Trial Standards for Civil and Commercial Law Cases Based on Network Flow Algorithm in Intelligent Judicial Environment

Qiang Zhang

Hongwei Liu

Haopeng Zhang

Publicado en línea: 29 sept 2025

Recibido: 21 ene 2025

Aceptado: 08 may 2025

DOI: https://doi.org/10.2478/amns-2025-1124

Palabras claveNetwork Flow Algorithm, Minimum Cost Flow, Base Feasible Graph, Smart Trial Construction

© 2025 Qiang Zhang, Hongwei Liu and Haopeng Zhang, published by Sciendo.

This work is licensed under the Creative Commons Attribution 4.0 International License.

Palabras clave
Network Flow Algorithm, Minimum Cost Flow, Base Feasible Graph, Smart Trial Construction