Analysis of Strategies to Combat Cargo Theft and Robbery in Peripheral Communities of São Paulo, Brazil, Using a Paraconsistent Expert System

Abstract

Background: Cargo theft represents a persistent challenge to last-mile logistics in the peripheral regions of São Paulo, Brazil, compromising transportation security and increasing operational costs. These high-crime areas disrupt supply chain stability and hinder e-commerce growth. Traditional security methods often fail to address the complexity and uncertainty present in these environments, necessitating adaptive approaches. Methods: This study applies an Expert System based on Paraconsistent Annotated Evidential Logic Eτ to assess the effectiveness of security interventions. Logic Eτ is particularly suited for analyzing uncertain, incomplete, and contradictory data in complex logistics settings. A mixed-methods approach was employed, integrating evaluations from nine experts representing different hierarchical levels within a logistics company. Six key security measures, including GPS tracking, armed escorts, optimized delivery windows, and the hiring of local drivers, were analyzed using favorable degrees and unfavorable degrees for each parameter. Results: The results demonstrated that five measures were effective, contributing to a 58% reduction in security costs in Arujá and 75% in Cajamar, two major distribution hubs. Conclusions: This study highlights the potential of combining Expert Systems and Eτ Logic to enhance cargo transport security, offering a scalable decision support framework for companies operating in high-risk urban regions.

1. Introduction

Peripheral communities, often overlooked by economic groups and businesses in Brazil, serve as significant hubs of life, culture, and economic activity. These regions, home to approximately 17.1 million people on the outskirts of Brazilian cities [1], generate an estimated annual revenue of BRL 119.8 billion, underscoring their substantial contribution to the national economy [2]. Recent studies indicate that 39% of residents in these areas engage in e-commerce, with a predominance of young consumers, of whom over 97% regularly access the internet, followed by 87% of adults using the web at least once per week [1]. These data reflect the increasing role of digital commerce in the economic inclusion of these communities, highlighting their growing demand for efficient and secure logistics operations [3].

Despite their economic potential, logistics operations in peripheral regions face significant structural and security-related challenges. Urban distribution, a critical obstacle for companies operating in emerging markets, is particularly complex due to poor urban planning, infrastructure, high congestion, non-standardized addressing, and elevated crime rates [4]. These logistical barriers severely hinder the integration of these communities into formal supply chains, restricting access to essential delivery services and perpetuating social inequality [5]. The lack of secure and structured logistics networks in these regions has led to operational inefficiencies, higher delivery costs, and increased vulnerability to crime. Recent research highlights that urban logistics requires innovative solutions to address these environments’ spatial and temporal variability characteristics [6]. Emerging technologies, including artificial intelligence-based decision-making models, have been proposed to address these challenges [7].

Cargo theft and robbery emerge as critical threats that compromise the security and efficiency of transportation networks. These crimes, particularly prevalent in peripheral regions of São Paulo, endanger logistics companies’ physical and financial stability and contribute to broader socio-economic instability [5]. Such incidents lead to increased operational costs, delays, and a diminished consumer experience, ultimately affecting the competitiveness of businesses in the e-commerce sector [8]. Transport companies have responded by investing significantly in security measures, including armed escorts, advanced tracking technologies, and optimized route planning. However, despite these efforts, cargo theft remains a persistent problem, demonstrating the need for more adaptive and effective strategies [9]. The unpredictability of criminal activity and the dynamic nature of logistics operations require decision-making models that can process large volumes of data while handling uncertainty and contradictions in real time [5].

Traditional security and logistics planning methods often rely on historical data and deterministic approaches that fail to account for real-time variability in crime patterns and logistical disruptions. In high-risk environments, adapting to constantly changing conditions is crucial. Decision-making in such contexts must incorporate methodologies capable of dealing with uncertainty, conflicting information, and incomplete data. In this regard, artificial intelligence-based solutions, notably Expert Systems integrated with advanced logical reasoning models, provide a promising approach to addressing these challenges [10]. By utilizing artificial intelligence, companies can enhance decision-making by processing large and ambiguous datasets, generating dynamic security strategies, and identifying emerging threats before they escalate into significant losses [11].

In response to these challenges, this study proposes using an Expert System based on Paraconsistent Logic Eτ to improve cargo security in peripheral regions of São Paulo. Paraconsistent Logic Eτ enables the analysis of contradictory and uncertain data, an inherent characteristic of complex logistics environments, allowing decision-makers to assess risks more accurately and optimize security interventions [12,13]. This approach enhances real-time adaptability by continuously refining the evaluation of security threats and improving operational responses. Unlike traditional binary logic, which forces decisions into rigid true or false classifications, Paraconsistent Logic Eτ accommodates varying degrees of uncertainty, enabling more nuanced and intelligent risk management in logistics [14,15].

This research aims to develop and evaluate innovative strategies for reducing cargo theft and robbery in São Paulo’s peripheral regions by leveraging an Expert System with Paraconsistent Logic Eτ. By integrating artificial intelligence with logistics security, this study aims to provide valuable insights into applying intelligent decision support systems in mitigating logistics risks. The proposed approach presents a replicable and scalable model that can be adapted to other high-risk urban areas, contributing to developing more resilient and efficient logistics networks. In addition to offering practical implications for logistics companies, this study aligns with broader urban security strategies and policy discussions, providing a framework for integrating technological solutions into crime prevention and supply chain resilience.

2. Background

2.1. The Seriousness of the Crime Scenario Related to Cargo Theft and Its Adverse Impacts on Transport Logistics

The high rates of cargo theft in São Paulo, Brazil, represent a critical crisis that demands immediate attention. The severity of this issue is evident in the significant number of incidents recorded by logistics companies, which directly affect both public safety and the operational efficiency of transportation networks [5,16]. Cargo theft is not merely a logistical inconvenience; it jeopardizes economic stability, disrupts supply chains, and diminishes confidence in the transportation industry.

Cargo theft has become a recurring issue in São Paulo, leading to substantial financial losses for transporters, insurance companies, and suppliers [4,9]. Beyond the direct loss of goods, these crimes generate additional expenses related to operational disruptions, criminal investigations, and the need to implement more robust security measures. These factors negatively affect business competitiveness, increase costs for end consumers, and damage the reputation of affected companies [7,16].

Additionally, rising operational costs from these security threats necessitate increased investments in armed escorts, GPS tracking technologies, and cargo insurance, further straining logistics budgets [5]. This challenge is particularly acute in highly competitive markets, where maintaining operational margins is already a significant concern [17].

As illustrated in

Table 1. Total losses from theft and robbery in 2023 by region.

Total Losses from Robbery and Theft in 2023	Center	East Zone	North Zone	West Zone	South Zone	Grand Total
Jan.			BRL 19,205.60	BRL 13,076.32	BRL 11,345.69	BRL 43,627.61
Feb.		BRL 5466.11			BRL 13,499.66	BRL 18,965.77
Apr.		BRL 30,857.68		BRL 18,650.09	BRL 3951.48	BRL 53,459.25
May		BRL 738.59				BRL 738.59
Jun.		BRL 682.93		BRL 8242.42		BRL 8925.35
Jul.			BRL 1509.59	BRL 479.06	BRL 4840.71	BRL 6829.36
Aug.		BRL 9454.38		BRL 7527.09	BRL 26,179.55	BRL 43,161.02
Sep.	BRL 1104.26	BRL 22,034.53	BRL 1407.73	BRL 1223.43	BRL 7604.88	BRL 33,374.83
Oct.	BRL 788.84	BRL 21,904.10	BRL 7617.46		BRL 9016.53	BRL 39,326.93
Nov.		BRL 31,127.77				BRL 31,127.77
Grand Total	BRL 1893.10	BRL 122,266.09	BRL 29,740.38	BRL 49,198.41	BRL 76,438.50	BRL 279,536.48

Source: Authors’ elaboration.

, data collected in 2023 show that an e-commerce logistics company recorded total losses of BRL 279,536.48. The eastern region of Greater São Paulo accounted for most of these losses, followed by the southern region, reinforcing the need for targeted strategies to mitigate risks associated with cargo transport in high-crime areas.

Logistics companies must invest in increasingly sophisticated security solutions to protect their cargo, including hiring private security teams, implementing advanced tracking systems, and reinforcing vehicle security [5,17]. However, despite these efforts, cargo theft remains a significant obstacle, affecting company profitability, competitiveness, and customer satisfaction due to delays and supply chain disruptions [4,7].

Furthermore, cargo theft places transport workers at high risk, with truck drivers and logistics personnel facing physical violence during theft incidents. The psychological toll on these professionals is substantial as they navigate constant threats to their personal safety and emotional well-being [5,9].

Amidst these challenges, São Paulo’s high rates of cargo theft pose broader economic threats. Beyond business losses, these crimes contribute to public perceptions of insecurity, which can deter investors and negatively affect economic growth [16,18]. A more effective, technology-driven security approach is essential to combat these challenges and enhance the resilience of logistics operations.

2.2. Paraconsistent Annotated Evidential Logic Eτ—Logic Eτ

Logic Eτ is a non-classical logical system belonging to the paraconsistent and paracomplete logics family. The Logic Eτ can serve as the underlying logic for inconsistent but non-trivial theories, i.e., rejects the Non-Contradiction Principle. Thus, such a logic category allows contradictory propositions (p and ¬p (the negation of p)) to coexist without rendering the system meaningless. Conversely, a paracomplete logic can serve as the foundation for theories where both a proposition (p) and its negation (¬p) are false, thereby rejecting the Principle of the Excluded Middle [19,20]. It is observed that the notion of paraconsistency and paracompleteness are independent.

Logic Eτ integrates paraconsistent and paracomplete characteristics, making it a highly adaptable framework for analyzing uncertain, contradictory, and complex data, a crucial requirement in logistics security analysis.

In Logic Eτ, atomic propositions are defined as p(μ, λ), where

• μ represents the degree of favorable evidence for a given proposition p.
• λ represents the degree of contrary evidence of the proposition p.

These values are defined on a closed real interval [0, 1], allowing for graded truth values rather than binary classifications. Thus, it enables a nuanced interpretation of data, such as the following:

• p(1.0, 0.0) → can be read as a true proposition (total favorable evidence, no unfavorable evidence).
• p(0.0, 1.0) → can be read as a false proposition (no favorable evidence, total unfavorable evidence).
• p(1.0, 1.0) → can be read as an inconsistent proposition (total favorable and unfavorable evidence).
• p(0.0, 0.0) → can be read as a paracomplete (complete lack of evidences).

The certainty and uncertainty degrees in Logic Eτ are defined as follows:

• Degree of certainty: Dce(μ, λ) = μ − λ.
• Degree of uncertainty: Dun(μ, λ) = μ + λ − 1.

As shown in

Table 2. Extreme logical states.

Extreme States	Symbol
True	V
False	F
Inconsistent	T
Paracomplete	⊥

Source: [14].

, the values indicate the extreme logical states.

Table 3. Non-extreme logical states.

Non-Extreme States	Symbol
Quasi-true tending to Inconsistent	QV→T
Quasi-true tending to Paracomplete	QV→⊥
Quasi-false tending to Inconsistent	QF→T
Quasi-false tending to Paracomplete	QF→⊥
Quasi-Inconsistent tending to True	QT→V
Quasi-Inconsistent tending to False	QT→F
Quasi-Paracomplete tending to True	Q⊥→V
Quasi-Paracomplete tending to False	Q⊥→F

Source: [14].

indicates the non-extreme logical states.

These properties allow Logic Eτ to manage uncertain, contradictory, and paracomplete data without leading to logical collapse, making it particularly suitable for decision-making models in security-sensitive logistics environments [12,14,15].

Figure 1 and Figure 2 illustrate how uncertainty and certainty degrees are represented in the decision lattice, providing a visual framework for analyzing logistics security risks.

In Figure 1, the values indicate the extreme and non-extreme decision states within the lattice.

As shown in Figure 2, the values in the diagram indicate the degrees of uncertainty and certainty, with adjustable limit control values indicated on the axes.

Consider a situation in which a driver reports feeling unsafe while delivering in a high-crime neighborhood, suggesting the presence of suspicious individuals. Simultaneously, the GPS tracking system shows that the vehicle follows the planned route without deviations and the delivery proceeds on schedule. Under traditional binary logic systems, this conflicting information might lead to a binary evaluation: either the delivery is proceeding safely (trusting the GPS data) or at risk (trusting the driver’s perception).

However, Paraconsistent Logic Eτ allows both pieces of evidence to be considered simultaneously, assigning a degree of favorable evidence to the GPS data (e.g., 0.9) and a degree of unfavorable evidence to the driver’s risk perception (e.g., 0.7). The system calculates the degrees of certainty and uncertainty, determining that while the GPS suggests normality, the driver’s concern warrants additional attention. Such information allows the company to adopt a cautious approach, such as contacting the driver for additional context or dispatching a nearby security escort, without prematurely halting the delivery. This flexible evaluation prevents false alarms while ensuring that potential risks are not ignored.

2.3. The Use of Expert Systems in the Context of Cargo Transport Security

Expert Systems (ESs) represent a powerful tool for addressing high-risk logistics challenges, such as cargo theft and robbery. These systems integrate specialized knowledge, heuristics, and rule-based decision-making models to support real-time risk assessment and security strategy development [12,15].

An Expert System’s architecture comprises three main components:

1.

Knowledge Base:

• Stores security protocols, past theft incidents, risk mitigation strategies, and expert heuristics.

2.

Inference Mechanism:

• Processes stored knowledge to generate risk assessments and recommendations using Paraconsistent Logic Eτ to handle contradictory or incomplete data [14,20].

3.

User Interface:

• Allows logistics managers to interact with the system, input real-time data, and receive strategic security insights [13].

Paraconsistent Logic Eτ plays a critical role in the ES’s inference engine, enabling the system to do the following:

• Analyze uncertain or conflicting crime data.
• Improve security decision-making accuracy by identifying crime patterns.
• Support adaptive security planning based on real-time risk fluctuations [14,21].

Furthermore, logistics professionals such as security managers and transportation analysts can integrate their subject matter expertise with the Expert System, refining risk assessments and validating preventive strategies [19].

By leveraging AI-driven decision support, logistics companies can enhance security protocols, optimize risk mitigation, and increase operational resilience against cargo theft [12,21]. This data-driven approach strengthens the industry’s ability to develop more intelligent, effective countermeasures against criminal activities targeting cargo transport.

3. Materials and Methods

3.1. Architecture and Functionalities of the Paraconsistent Expert System

Applying Expert Systems in cargo transport security requires an integrative and multidisciplinary approach incorporating technical principles, advanced analytical strategies, and robust computational tools. This approach ensures the development and implementation of preventive and intervention measures that are both effective and adaptable to high-risk and complex logistical environments [12,13]. Given the increasing frequency of cargo theft and the uncertainties inherent in security decision-making, using intelligent decision support systems based on Paraconsistent Logic Eτ offers a viable and scalable alternative for mitigating these challenges.

Expert Systems play a crucial role in analyzing and solving complex problems in logistics and security, where decision-making often relies on uncertain and contradictory data [14,15]. Historical data are collected from company internal reports, police records, and insurance claims, ensuring a comprehensive view of cargo theft incidents. These data are validated by cross-checking these sources, eliminating discrepancies and enhancing reliability. Additionally, the system integrates data from external sensors, including GPS devices for geolocation, vehicle telemetry systems for route deviations and driving behavior, and RFID tags for cargo identification. These systems replicate the decision-making abilities of human experts by integrating a structured knowledge base, inference mechanisms, and interactive interfaces. By doing so, they enhance the ability to make real-time security decisions, even under conditions of ambiguity and incomplete information.

The core components of the Paraconsistent Expert System include the following:

1.

Knowledge Base:

• This repository stores specialized security knowledge, including historical data on cargo theft, robbery trends, security regulations, risk mitigation strategies, and expert heuristics. These elements provide a foundation for logical reasoning and generate contextualized security solutions [13,14].

2.

Inference Engine:

• Acting as the “brain” of the system, the inference engine processes and interprets information from the knowledge base. It utilizes advanced logical reasoning algorithms, including those based on Paraconsistent Logic Eτ, to analyze security data and resolve conflicting information [13,16].

3.

User Interface:

• This interface facilitates operator interaction with the system, allowing users to input data, access security insights, and visualize risk analysis. Depending on operational requirements, this interface may range from textual command line interfaces to sophisticated graphical dashboards for logistics professionals [5].

The system’s applicability extends beyond problem identification, as it facilitates real-time decision-making, enhances risk assessment, and assists in defining long-term security policies. By integrating Logic Eτ, the system ensures that optimal decisions can still be made even in conflicting intelligence reports or uncertain risk levels [4,12].

3.2. Application of the Paraconsistent Expert System in Cargo Transport Security

The Expert System is applied in various ways to enhance risk mitigation, strategic planning, and real-time security adaptation in cargo transportation security.

The first core function is risk analysis, which uses historical crime data and real-time tracking inputs to identify theft-prone zones. The system employs predictive analytics to assess vulnerabilities and recommend preventive measures based on detected trends [5,9].

The second application is decision-making, where the system optimizes transport routes, suggests secure delivery windows, and ensures efficient resource allocation. Given that theft patterns evolve, the ability of the system to handle contradictory risk assessments ensures that the most reliable decision is made [15,16].

Lastly, the strategic planning function assists in long-term security management. The system enables transport companies to proactively invest in security measures, such as cargo tracking infrastructure, secure parking zones, and personnel training, based on evolving crime intelligence [4,5].

By implementing this multi-layered approach, the Paraconsistent Expert System provides cargo carriers with an adaptive and responsive framework, ensuring logistics security remains resilient even in volatile environments.

3.3. The Application of Paraconsistent Logic Eτ in Data Analysis

Paraconsistent Logic Eτ is a non-classical logic framework that manages imprecise, contradictory, and ambiguous data [12,15]. In cargo security, this logic is a powerful tool for handling conflicting intelligence reports, fluctuating risk levels, and fragmented crime data.

The first fundamental function is handling uncertainty. Traditional binary logic fails in complex logistics settings where risk is not absolute. Paraconsistent Logic Eτ allows for gradual assessments rather than absolute truths, making it ideal for dynamic security monitoring [9,16].

The second application is inconsistency resolution, where the logic framework detects contradictions in security intelligence. For example, when different sources report conflicting crime levels in the same region, the logic model reconciles these inputs and prioritizes the most reliable outcome.

The final function is an interpretation of ambiguous intelligence. Crime-related reports are often fragmented, subjective, or open to multiple interpretations. The Logic Eτ enables a structured approach to analyzing unclear security reports, preventing biased decision-making due to incomplete threat assessments [4,5].

By integrating this framework, logistics operators and law enforcement agencies can develop adaptive cargo security models, ensuring continuous refinement of security protocols based on evolving risk landscapes.

3.4. Evaluation of Security Actions Implemented in Cargo Transport

A structured evaluation process was developed to assess the effectiveness of cargo security measures, integrating qualitative and quantitative methodologies [13,14].

One of the first initiatives was GPS-based monitoring, which enabled real-time tracking and automated anomaly detection, allowing for instant intervention in case of unauthorized cargo movement [5,9].

A second measure was considered, and it involved deploying armed escorts for cargo shipments along high-risk routes. The impact of this intervention was evaluated in terms of its deterrence effect on criminal activities and its success in preventing theft incidents [4].

Additional security strategies included defining restricted delivery windows, limiting cargo movement during high-crime hours, and significantly reducing exposure to theft-prone conditions [5].

Integrating these measures within the Paraconsistent Expert System allowed for dynamic adaptation. Security protocols were continuously updated based on real-time intelligence, ensuring progressive refinement of cargo protection strategies [9].

3.5. Expert Analysis and Decision-Making

Structured expert evaluations were conducted to validate the implemented security measures. Three groups of logistics professionals were consulted to analyze and assess the effectiveness of various security interventions [4,14].

The selection of experts followed a structured process to ensure a diverse and knowledgeable representation of professionals engaged in cargo transport security. The criteria for selection included (1) professionals with at least five years of experience in logistics operations or security and logistics operations management, (2) individuals holding managerial or executive positions responsible for strategic decision-making in transport security, and (3) frontline personnel with direct exposure to security risks, such as drivers and delivery people. Experts were chosen based on their extensive field experience, operational insights, and strategic perspectives to comprehensively evaluate security measures’ effectiveness. This methodology aligns with best practices in expert-based decision-making for risk assessment and security strategy validation [9,16].

This multi-tiered approach aligns with best practices in logistics operations research, where capturing insights from strategic decision-makers and field operators is essential for developing holistic risk mitigation strategies [5,9].

The sample size of nine experts was determined based on a purposive sampling strategy, commonly used in expert-driven qualitative research where the focus is on depth rather than breadth [20]. Given the specialized nature of the study, the primary objective was to gather insights from professionals with direct experience in cargo transportation security. Prior studies on expert evaluations in logistics and security management have demonstrated that a sample size of between 6 and 12 participants is sufficient to achieve data saturation, where additional responses do not significantly alter the findings [9,22,23]. Moreover, the Paraconsistent Logic Eτ methodology ensures that even with a limited sample size, conflicting data can be systematically resolved and analyzed to derive meaningful insights [14].

The division into three groups was designed to capture distinct perspectives across the logistics security chain. Logistics executives provide high-level strategic insights regarding corporate risk management and investment decisions in security technologies. Operational coordinators offer an understanding of the practical implementation of security protocols and logistical constraints. Frontline professionals, including drivers and delivery people, provide firsthand experience on the real-world effectiveness of safety interventions in high-risk transport routes. This segmentation ensures a holistic analysis considering strategic decision-making and operational execution challenges [5,16,24].

Experts were categorized into three main groups:

1.

Logistics Executives:

• Responsible for assessing strategic security implications within corporate frameworks.

2.

Operational Coordinators:

• Evaluated how security interventions affected day-to-day logistics operations and risk mitigation.

3.

Frontline Professionals (Drivers and Delivery people):

• Provided direct field insights on the practicality and effectiveness of implemented measures [5].

A Google Forms survey was developed to collect structured expert feedback. Respondents assigned degrees of certainty and uncertainty to specific security measures, quantifying perceived effectiveness [9].

The structured survey consisted of six key statements, each evaluated twice, once for certainty and once for uncertainty, resulting in 12 response points per expert. Responses were collected using a Likert scale from 1 to 5, where 1 indicated strong disagreement and 5 indicated strong agreement. The data were then processed using Paraconsistent Logic Eτ, which allowed for the resolution of conflicting responses. The algorithm assigned confidence scores based on the degree of consensus among experts, weighting the certainty and uncertainty levels accordingly. The results ensured that final evaluations reflected expert agreement and potential data inconsistencies, optimizing the decision-making framework [4,14,25].

Figure 3 shows how the Forms form was applied to three groups of experts who are knowledge holders in the area to analyze the actions proposed in 2023 and implemented in 2024, using statements in which the experts attributed degrees of certainty and uncertainty to each of the statements:

The participants read the statements twice. During the first reading, they chose between Strongly Agree and Agree on a Likert scale (1 to 5). They chose between Strongly Disagree and Disagree during the second reading on a Likert scale (1 to 5).

The statements made in the questionnaire were as follows:

• Implementing Escorts on the most critical routes for the inhibition of claims is effective.
• Reducing the frequency of service in the most critical regions to reduce exposure of cargo is effective.
• Creating specific delivery windows reduces the risk of cargo theft and robbery.
• The search for drivers who live in critical regions to reduce the risk of accidents due to the driver’s full knowledge of the region reduces the risks.
• The loyalty of drivers and helpers in critical areas to retain their knowledge of risk zones and local movement reduces the risk of accidents.
• Monitoring loads using GPS to identify anomalies is an effective action.

The forms applied to the experts contain six statements, each of which was repeated so that in one, the expert could assign it a degree of certainty and, in another, a degree of uncertainty. That means that each interviewee had 12 options to analyze.

In order to define the groups of interviewees, it was taken into account that a predominant group of executives at the managerial level would be needed to analyze the impacts according to the strategic planning of the company being researched, a group of specialists in coordinating transport operations would also be needed in order to analyze the impacts that the actions would have within the distribution center, and finally, a group of frontline delivery professionals in risk areas such as the outskirts of São Paulo made up of drivers and helpers would be required as well.

Table 4. Expert groups and their representatives.

Groups	Position of the Specialist Interviewed
Group A	Logistics General Manager
	Logistics Senior Manager
	Logistics Operations Manager
Group B	Logistics Operations Coordinator
	Logistics Operations Supervisor Logistics Operations Supervisor
Group C	Driver/Delivery person (Critical Regions)
	Driver/Delivery person (Critical Regions) Driver/Delivery person (Critical Regions)

Source: Authors’ elaboration.

lists the groups and hierarchical levels of the professionals who took part in the proposed analysis.

The collected responses were processed using Logic Eτ, ensuring that conflicting expert opinions were resolved within a structured analytical model [4,14].

Experts from distinct hierarchical levels—strategic executives, operational coordinators, and frontline drivers—were purposefully selected to address potential biases and enhance the validity of the evaluations. This multi-level approach ensured a comprehensive view of the security measures’ effectiveness, capturing both strategic and operational perspectives. Additionally, the application of Paraconsistent Logic Eτ further mitigated individual subjectivity by balancing favorable and unfavorable evidence degrees, allowing the system to handle conflicting assessments and reduce the impact of personal bias on final decisions.

Following the application of Logic Eτ, the results were categorized into four states: true, false, inconsistent, and paracomplete. The final decision on each security intervention was determined by assessing the degree of certainty and uncertainty derived from expert evaluations. Actions with a certainty score above the predefined threshold (e.g., 0.5) were classified as effective interventions, whereas those with high uncertainty scores required further calibration. This statistical categorization provided a structured basis for refining security strategies and optimizing future cargo protection measures [9,15,26].

By utilizing expert-driven evaluations, real-time crime intelligence, and Logic Eτ based decision models, this study established a scalable framework for ongoing refinement of cargo security strategies.

4. Results

Database Construction

To support the application of Logic Eτ, a structured database was developed using expert knowledge to assess the effectiveness of the proposed security measures. Nine experts were selected and divided into three groups according to their professional expertise. These experts evaluated each security action, assigning degrees of favorable evidence (ai R) and degrees of unfavorable evidence (bi R), which were systematically compiled into the dataset presented in

Table 5. Dataset containing expert assessments of each analyzed factor.

	Group A						Group B						Group C
Factor	Expert 1		Expert 2		Expert 3		Expert 4		Expert 5		Expert 6		Expert 7		Expert 8		Expert 9
	ai,1	bi,1	ai,2	bi,2	ai,3	bi,3	ai,4	bi,4	ai,5	bi,5	ai,6	bi,6	ai,7	bi,7	ai,8	bi,8	ai,9	bi,9
F01	0.20	1.00	0.20	1.00	1.00	0.20	0.60	0.20	0.80	0.60	1.00	0.40	0.60	0.60	1.00	0.00	0.80	0.60
F02	0.20	1.00	0.60	0.20	0.60	0.20	0.20	1.00	0.80	0.60	0.40	0.00	0.60	1.00	0.60	0.40	0.60	1.00
F03	0.80	0.40	0.80	0.20	1.00	0.20	0.40	0.60	0.40	1.00	1.00	0.40	1.00	0.60	1.00	0.00	0.60	1.00
F04	0.80	0.40	1.00	0.20	1.00	0.20	1.00	0.20	1.00	0.20	0.80	0.40	0.60	0.60	1.00	0.00	1.00	0.00
F05	0.80	0.20	0.80	0.20	1.00	0.20	1.00	0.20	1.00	0.20	1.00	0.40	1.00	0.00	1.00	0.00	1.00	0.00
F06	0.80	0.20	0.60	0.40	1.00	0.20	1.00	0.20	1.00	0.20	0.80	0.20	0.40	0.80	1.00	0.00	1.00	0.00

Source: Authors’ elaboration.

.

At this stage, it is necessary to assign weights to each expert to reflect their expertise and knowledge regarding the security measures evaluated. In this study, an equal weight was assigned to all experts to maintain analytical neutrality. However, weight calibration may be introduced in future iterations to better differentiate expert influence based on their experience and field knowledge.

Expert opinions were logically processed using the maximization and minimization rules, ensuring that contradictory and uncertain information was systematically addressed. The following logical operations were applied, as per Section 2.2 (Application of Logic Eτ):

[(expert 1) OR (expert 2) OR (expert 3) AND (expert 2) OR (expert 5) OR (expert 6)]

This approach ensures that the most supported and least contradictory opinions prevail, offering a structured decision framework. The outcomes of these calculations are presented in

Table 6. Results obtained from applying maximization and minimization rules and the corresponding weighted arithmetic mean of the resulting degrees.

	A		B		C		Resulting Degrees
Factor	MAX [E1, E2, E3]		MAX [E4, E5, E6]		MAX [E7, E8, E9]		MIN {A, B, C}
	ai,Ga	bi,gA	ai,gB	bi,gB	ai,gC	bi,gC	a1,R	b1,R
F01	1.00	0.20	1.00	0.20	1.00	0.00	1.00	0.20
F02	0.60	0.20	0.80	0.00	0.60	0.40	0.60	0.40
F03	1.00	0.20	1.00	0.40	1.00	0.00	1.00	0.40
F04	1.00	0.20	1.00	0.20	1.00	0.00	1.00	0.20
F05	1.00	0.20	1.00	0.20	1.00	0.00	1.00	0.20
F06	1.00	0.20	1.00	0.20	1.00	0.00	1.00	0.20
	Overall Analysis: Average of Resulting Degrees						0.933	0.267

Source: Authors’ elaboration.

.

Following this step, the degree of certainty (Dcer) and degree of uncertainty (Dun) were calculated to determine the logical state of each evaluated proposition. The following results were obtained:

Dcer (Pi × ai, R; Pi × bi, R) = ∣5.6/6∣ = 0.933 (93%)

Dun (Pi × ai, R; Pi × bi, R) = ∣1.6/6∣ = 0.267 (26.7%)

In Logic Eτ, the determination of external and arbitrary limit values is crucial to define whether a proposition should be classified as true, false, inconsistent, or paracomplete. In this study, the preliminary limit values were set as follows:

Vnear true = 0.5

Vnear false = −0.5

Vnear inconsistent = 0.5

Vnear paracomplete = −0.5

These threshold values act as reference points for decision classification. However, in future studies, these limits may require further calibration to enhance accuracy based on additional expert evaluations and real-world operational data.

The logical states for each proposition were computed and classified using the parameters according to Logic Eτ rules. The results are presented in

Table 7. Classification of resulting states according to Logic Eτ.

	Conclusions			Weighting of the Resulting Degrees
Factor	H	G	Decision	Pi × ai,R	Pi × bi,R
F01	0.80	0.20	TRUE	1.0	0.2
F02	0.20	0.00	NOT CONCLUSIVE	0.6	0.4
F03	0.60	0.40	TRUE	1.0	0.4
F04	0.80	0.20	TRUE	1.0	0.2
F05	0.80	0.20	TRUE	1.0	0.2
F06	0.80	0.20	TRUE	1.0	0.2
	0.667	0.200	TRUE	5.6	1.6

Source: Authors’ elaboration.

, summarizing the outcome of each evaluated security action.

The analysis confirms that most of the evaluated security actions exhibit high degrees of certainty, reinforcing their effectiveness in mitigating cargo theft risks. However, specific factors were identified with notable uncertainty levels, indicating the need for further refinement and additional control measures in future iterations of the security framework.

5. Discussion

The results obtained in this study highlight the efficiency of the strategies implemented to mitigate cargo theft and robbery in São Paulo’s peripheral areas. These findings were derived from structured analyses conducted using the Expert System (ES) based on Logic Eτ, which processed qualitative and quantitative data collected from experts at different hierarchical levels of the organization. The Logic Eτ methodology allowed for the assessment of uncertain and contradictory expert opinions, enabling a refined decision-making process that increased the reliability of the conclusions drawn from this research [13,14].

The study determined that five out of six actions implemented by the company resulted in a viable final decision, whereas one action remained inconclusive. However, further analysis revealed a high probability that this action could become fully viable, depending on how expert uncertainty levels evolve. This outcome underscores the strength of Logic Eτ in handling decision-making under uncertainty, as it enables robust conclusions despite inconsistencies in expert opinions, variations in experience levels, and complex operational factors [20,21,27].

Figure 4 illustrates the financial impact of these security measures, presenting the consolidated investments in cargo security by an e-commerce company specializing in furniture deliveries.

A crucial finding of this research is that after implementing targeted security measures, combined with assessments conducted through the Logic Eτ Expert System, the company experienced a significant reduction in security-related expenses. Specifically, security escort costs decreased by approximately 58% in Arujá and 75% in Cajamar, two major distribution centers in São Paulo’s metropolitan area. These cost reductions have substantial financial implications, demonstrating the practical viability and economic efficiency of adopting intelligent expert systems for logistics security management.

Additionally, interviews with field experts identified process bottlenecks and unexpected risk factors affecting cargo transportation security. Among the most critical issues were hiring security escorts with limited experience in high-crime areas and the long-term retention of employees assigned to high-risk routes. In some cases, loyalty to specific routes led to security vulnerabilities, as employees could develop collaborations with organized crime groups involved in cargo theft in São Paulo’s peripheral communities [5,16].

Additional predictive countermeasures are recommended to refine security protocols and prevent unintended insider threats. These include recruiting specialized escort teams with prior experience in high-crime zones and adopting more rigorous hiring and background check policies to mitigate the risks posed by collusion between transport workers and criminal networks.

The Para-Analyzer device was used as the final data validation stage, as depicted in Figure 5. This device facilitated a structured, quantitative assessment of expert judgments, consolidating results into an objective framework for further strategic decision-making [12,15].

The findings indicate that Logic Eτ based analysis, when applied to expert-driven security decisions, significantly enhances the effectiveness of risk management strategies. The insights from this study contribute to advancing security-focused AI applications in last-mile logistics, with the potential to influence future urban freight transport policies and industrial security protocols worldwide.

6. Conclusions

This study has provided a robust analytical framework for applying Logic Eτ in security-sensitive logistics operations, demonstrating its effectiveness in processing uncertain and contradictory information. Cargo transport security is inherently volatile, with real-time risk factors that challenge traditional decision-making models. Logic Eτ based Expert Systems emerge as a highly effective alternative, capable of mitigating uncertainty, optimizing security measures, and enhancing operational efficiency [14,20,24].

A significant contribution of this study is its demonstration of how Expert Systems analyze collected data and continuously learn from new security incidents. This adaptive learning capability ensures that as new patterns of cargo theft emerge, the system dynamically adjusts its security strategies, thereby proactively preventing risks rather than reacting to threats after they occur [13,15].

Additionally, the findings reinforce the economic benefits of deploying intelligent decision support systems in logistics security. The cost reductions observed in security-related investments in São Paulo’s Arujá and Cajamar distribution centers highlight the financial viability of these AI-driven security models. The results confirm that technology-based interventions are effective and cost-efficient, making them an attractive solution for e-commerce logistics providers operating in high-risk urban areas.

Ultimately, integrating a Logic Eτ-based Expert System into logistics security strategies represents a major advancement in AI-driven decision-making. By combining human expertise with AI-powered analytical models, companies can achieve higher precision in security management, reducing cargo theft risks while protecting supply chain integrity. Given the escalating security challenges in urban logistics worldwide, the findings from this study provide a replicable framework for improving cargo transport security in high-risk regions [5,12,26].

7. Suggestions for Future Work

This study’s primary objective was to evaluate the effectiveness of an Expert System based on Logic Eτ in mitigating cargo theft and robbery in São Paulo’s peripheral regions. While the current findings demonstrate significant progress, there remains room for further refinement and expansion. Several recommendations are proposed for future studies and practical implementations, including the following:

• Enhancing Security Escort Teams: Rehiring highly experienced escort professionals who specialize in navigating high-crime areas.
• Integrating Advanced Surveillance Technologies: Deploying side- and rear-mounted truck cameras connected to a centralized monitoring system that can detect suspicious activities in real time.
• Implementing Rigorous Personnel Screening: Conducting criminal background checks for all current and prospective drivers, ensuring they have no prior affiliations with cargo theft networks.
• Expanding the Scope of the Expert System: Optimizing decision-making capabilities of the Paraconsistent Expert System to include machine-learning algorithms, which would allow for automated, adaptive security improvements.

These measures would further enhance logistics security, ensure safer cargo transport, and increase profitability, making the Logic Eτ Expert System a key technological asset for urban freight transport security management.