Automation of Negative Infrastructural Externalities Assessment Methods to Determine the Cost of Land Resources Based on the Development of a “Thin Client” Model

Abstract

This article discusses the need to transform real estate valuation methods. It is associated with the problems of obtaining unreliable results affecting the subsequent adoption of management decisions. As an important element of land plots assessment, the authors define the Negative Infrastructural Externalities arising from the presence of infrastructure and other regime-forming facilities. These externalities represent the loss of title holders due to the encumbrances arising from the use of land plots. The world community (and the authors as part of it) sees one of the transformation methods in the automation of the evaluation process. Therefore, the purpose of this study is to develop a mechanism of automating the Negative Infrastructural Externalities assessment process in the conditions of a non-existent and weak market activity of land relations. Modern trends dictate the saving of hardware, labor and money resources; in this connection, the methods of Negative Infrastructural Externalities assessment are implemented on the basis of the “thin client” technology. The research is based on the following methods: the analytical method is used to perform a critical analysis of the problem area and to substantiate the research topic’s relevance; methods of object-oriented programming and methods of modular programming (Cowan’s axiom of modularity) are used as tools for developing the web application logic, as well as the interaction of its individual elements; the attribute-driven design approach is used in the creation of software architectures. The result of the study is the developed and substantiated architecture of a web application for assessing negative infrastructural external factors in determining the land value, the implemented modular structure of the specified web application and the developed conceptual model of the database. The practical implementation of the listed proposals is made by means of the Python programming language. The advantage of the created automated system is the possibility of multi-disciplinary use of the expert assessment approach when changing the settings.

1. Introduction

The modern problem of sustainable development, outlined by the international declarations and strategic documents of numerous countries of the world, objectively affects various fields of citizens’ lives (Russia included). The land market, as a specific and priority socio-economic niche in Russia, is subject to special state regulation and “sustainable” management since the objects of land legal relations serve as the basis for the country’s spatial development and are also a national treasure and an irreplaceable means of production.

One of the most important levers of the regulation and intensification of land turnover is the assessment activity, which, following the transition to market relations at the end of the 20th century in Russia, became the most important instrument of the market economy, having received rapid development in the early stages. However, in the absence of a systematic approach to real estate valuation, there are still serious gaps in its legal, methodological and software support.

Valuation activity in Russia is gaining momentum and is carried out in accordance with international treaties—e.g., Federal Law No. 135-FZ of 29 July 1998 “On valuation activities in the Russian Federation”—as well as other regulatory legal acts. As an object of legal regulation, real estate valuation is an activity aimed at establishing a market or other value of valuation objects. In Russia, it is carried out taking into account the principles of evaluation and in accordance with the Federal Valuation Standards. The Standards define the main types of real estate value, indicate the need to apply three approaches to valuation and also set the period during which the value of the valuation object can be recommended for a real estate transaction. The professional activity of appraisers in Russia is currently regulated by the state and self-regulatory organizations.

The development of valuation activities in the Russian Federation today lies in the formation of a regulatory framework. There is a trend of focusing Russian valuation standards to international (ISO) and European (ECO) standards within the European Union for The European Group of Valuers’ Associations (TEGoVA). At the same time, given the passivity of the Russian real estate market, Russia is actively creating a national valuation system. In the current economic situation in Russia, the real estate market is imperfect due to high business activity in large cities of the European part of Russia and low market activity in the regions. On the other hand, no more than 7 years ago, the acquisition of prestigious land from the state was carried out as part of “shadow transactions”, when land plots that were expensive from the market point of view were sold at a low price. The created institutional infrastructure today has already laid a serious foundation for the development of market relations in the land sector. The main problem in assessing the market and cadastral value of real estate objects is the unreliability, disunity and inconsistency of the real estate market. For some segments and regions, the market for the sale of real estate is practically or completely absent, and this does not allow one to unambiguously assess the market value of real estate for hypothetical proposals in the absence of the differentiation of economic value and the presence of transaction costs.

The problems associated with real estate valuation are also highlighted by the world community. As noted in a work by Grover, R. [1], in the practice of the world community. there are serious problems associated with the type and quality of data used in mass valuation models, as well as the implementation of real estate valuation, including mass valuation.

Against the background of the innovative development of various industries, valuation activities in Russia, unfortunately, are (to some extent) in the stage of “stagnation”. This is primarily due to the blind use of western assessment methodologies without reference to the Russian market realities and to the specifics of the evaluated territories [2,3].

The need to develop methods and techniques for assessing real estate, including land, is increasing every year. There are new trends in solving the issues related to land and real estate taxation that appear against the background of the country’s socio-economic development; the preservation of historical and cultural heritage objects; the seizure of land for state and municipal needs for subsoil use and for the construction of engineering infrastructure; compensation for losses in connection with the limitation of the rights of title holders of real estate; solving social and environmental problems in urbanized areas; making decisions on investments in real estate; and intensifying the development of the real estate market as a whole and the adoption of government management decisions.

Researchers from many countries of the world have recently raised the issue of the need to transform the methods of real estate valuation in connection with the problems of obtaining high-quality results that may influence the decisions being made.

Yiorkas, Ch., and Dimopoulos, Th. [4] sharply raise the problem of the need to improve existing mass land valuation methods and propose more advanced mass land valuation methods based on the synthesis of GIS technologies, regression analysis and modeling interpretation using Geographically Weighted Regression (GWR). The authors pay special attention to the spatial attributes accounting that affects the value of real estate in the process of mass valuation.

With the increase in the volume of real estate transactions, the mass valuation of real estate, including land, is becoming widespread in the world [5]. At the same time, assessment methods are constantly being improved, and due to the growing influence of digitalization and computerization, machine learning has begun to play a special role in the process of creating mass assessment models. Lee, C. [6] has developed a hybrid approach to mass valuation based on a neural network for supervised learning and principal component analysis, which improves the reliability and validity of the assessment result.

Bogataj, M., Suban, Tuljak, D. and Drobne, S., in their work [7], proved that the use of a fuzzy model for determining the value of real estate improves the result by leveling the subjectivity of the appraiser and the fuzziness of pricing factors.

In 1996, Wyatt, P. [8] had substantiated the need for spatial analysis using the GIS in the process of real estate valuation with the aid of the comparison method. Gnat, S. [9] analyzed modern methods and models of mass land valuation, focusing on the need to obtain a large amount of data for this, which is not always possible. He researched machine learning algorithms and multiple regression models in mass real estate valuation in the low-activity market of Poland, having proved that, even when working with small datasets, accurate appraisal results can be expected thanks to machine learning algorithms (nonparametric models-nearest neighbors (KNN) regression and XGBoost) [9]. The effectiveness and advantages of machine learning models have also been confirmed in the South Korean real estate market by researchers such as Kim, Y., Choi, S., Yi, M.Y. [10]. Wang, D. and Li, V. [11], who identified the future trend of mass valuation as “mass valuation 2.0” or the AI trend: AI-based model, GIS-based mode, and MIX-based model. The authors note that the combination of the classical procedure for creating an assessment model and analyzing and testing an objects group for a given date with artificial intelligence is the future of mass valuation, Quantitative methods of mass assessment by Paloma, T., Kauko, T. and D’Amato, M. [12] are divided into four groups: expert methods, model-based methods, data-based methods and machine learning methods. Khamis, A., and Kamarudin, N.K., in their work [13], revealed the advantages of a neural network model over multiple regressions, which consist in reducing the error and improving the quality of the model. Baldominos, A., Blanco, I., Moreno, A. J., Iturrarte, R., Bernardez, O. and Afonso, C. [14] studied the problem of obtaining up-to-date information about real estate prices and developed an online application where, using machine learning algorithms, it is possible to predict the value of real estate. Dimopoulos, T. and Bakas, N. [15] developed four machine learning models that made it possible to obtain and analyze large amounts of data about the spatial characteristics of assessment objects, thereby increasing the accuracy of the result.

In modern conditions of the global trend towards the digitalization of various areas of knowledge and practice, automation is becoming the most relevant. The Real Estate Valuation Institute is no exception. Bennett, R.M., Unger, E.-M., Lemmen, C. and Dijkstra, P. [16] highlight the problem of a slow and sine-shaped transition to automated methods of land management and the automation of real estate valuation, developing models for the maintenance of land management systems. The development of market relations stimulated the development and application of automated evaluation models.

Automated Valuation Methods (AVM) combine computerized real estate databases and programming languages and are essential to ensuring the accuracy of a valuation. Such models are being developed by Schulz, R., Wersing, M. and Werwatz, A. [17], who note that automated outlier removal is important in building models and that AVM is designed to reduce the cost of evaluation. D. Demetriou notes that valuation is an important step in the process of agricultural land consolidation. At the same time, the author came to the conclusion that assessments performed manually and empirically without the use of systematic analytical tools lead to unreliable inflated values of the cost. The author has developed a linear hedonic model, which is based on spatial analysis and helps to reduce the evaluation time by 80% [18].

Droj, L. and Droj, G. [19] note that, in the process of assessment, the use of GIS allows for the obtention of accurate and objective information about the characteristics of the assessed land plots. Additionally, the spatial analysis of the territory helps the appraiser to substantiate the obtained result of the assessment, to automate some elements of the assessment and to obtain a more reliable result in a short time [6]. Glumac, B. and Des Rosiers, F. [20] introduced the concept of automated real estate and land valuation systems, described their advantages over automated models and developed their generalization and classification. The authors proposed their own taxonomy, which is not hierarchical, since all systems, in their opinion, are equivalent to each other. Arcuri, N., De Ruggiero, M., Salvo and F., Zinno, R. [21] suggest combining GIS and BIM into automated assessment methods. Among the studies dedicated to the development of real estate valuation models, one can highlight the study by Ciuna, M., Salvo, F. and Simonotti, M. [22], which presents the process of developing an AVM assessment model, characterized by the ability to perform even in the conditions with a small amount of data available; the article by Donald, E. [23], which describes the requirements for authorized assessment models; and the study by d’Amato, M. [24], which is dedicated to the use of automation for monitoring the assessment results. As a result, the development of automated modules for real estate valuation makes it possible to reduce labor and time costs, to obtain more accurate results and to operate with large amounts of data.

A sufficient number of scientific studies at the present stage of the development of assessment activity are dedicated to the improvement of methods of real estate valuation. However, when making such assessments, little attention has been paid to taking into account externalities (not compensated by the market).

The study of the negative impact of encumbrances in the use of land caused by overhead power lines (negative infrastructure externalities (NIE)) began in 1979 in the USA by Gustafson, R.J., Grumstnup, P.D., Herdrickson, E.R. and Meyer, M.P. [25]. Based on aerial photography of several states of America, the authors conducted a spatial analysis of the agricultural territories and established the negative impact of these lines on the productivity of crops, which leads to crop losses, time for harvesting and an increase in labor costs. Vaillencourt, F. and Monty, L. [26] identified the effect of regulation of agricultural land use on their prices using a regression model built on initial data, including more than 1200 undeveloped land plots. They found that burdened land sells for 15–30% less than unburdened land. In 1991, Patrick Beaton studied the effect of use restrictions on real estate values in New Jersey pine forests and found that prices for land use restrictions increased significantly from 1972 (when the restrictions were originally proposed) to 1981 (the year of the entry of the restrictions). Based on the results, W. Beaton suggested that the difference in price between more restricted and less restricted areas is more dependent on the strength of activity regulation. These studies also illustrate how pre-restriction development pressures can temporarily raise land prices in areas earmarked for future regulation [27].

According to Michael, J. and Palmquist, R. [28], some researchers, in order to explain the differences in the impact of land use regulation on the value of real estate, including land, in different settlements, apply restriction severity indices, counting their number and using them as an index in the hedonic regression model. Research by Ihlanfeldt, K.R. [29] has proven that land use regulation has a significant impact not only on the prices of undeveloped land but also on residential properties. The results suggest that estimates of these effects may be biased if the measure of regulatory constraint is seen as exogenous to housing prices or if the estimated effects cannot vary with market conditions.

Glaeser, E.L., Ponzetto, G.A., Zou, Y., Gyourko, J. and Saks, R. [30] consider externalities by analyzing the impact of the spatial structure of the urban economy on the welfare of the population. To assess the endogenous amenities of the city, it is necessary to take into account the exogenous effects that can provoke changes within the city. For example, to evaluate the impact of city size on overall amenities, researchers will need sources of exogenous variation and a structural model. To assess the impact of city amenities on the level of improvement, it is necessary to take into account the short-term or long-term perspective, the level of well-being of residents, pollution, traffic situation and some other factors. According to the authors, the ambiguity of amenities that arise in megacities can cause both an increase and decrease in the value of real estate, depending on the prospects for the use of territories [30,31]. Yin Huang, Tao Hong and Tao Ma compare the two concepts of urban networks and the agglomeration economy in terms of externality theory and then empirically study the impact of urban network externalities on city performance to determine which types of cities would benefit more from urban network externalities [32]. The possibility of the occurrence of negative externalities of production from the state infrastructure is considered by Boarnet, M.G. [33] for the example of the California districts data. Iranian scientists Najkar, N., Kohansal, M.R. and Ghorbani, M. [34] assess the impact of regional transport infrastructure externalities on agricultural productivity using the Darbin model, while Zhang, Yin-Fang and Shengbao, J. [35] assess the impact of infrastructure on regional industrial productivity in China and identify differences between intraregional and interregional externalities. An analysis of interregional externalities showed a positive short-term spatial interaction between the electric power industry and road infrastructure but not railways; in the long run, these infrastructure effects are positive among provinces that do not compete for production, but they become negative among competing provinces in terms of electricity and regions similar in terms of transport infrastructure [35].

In Russia, in the context of the significant differentiation (and, for the most part, depressiveness) of the real estate market activity, the following is of the utmost importance: the assessment of Negative Infrastructural Externalities (NIE), which arise in connection with the presence of infrastructure and other regime-forming objects. NIEs represent the losses of title holders encumbered with Zones with Special Territory Use Conditions (ZSTUC) of land plots caused by operational restrictions, spatial consequences, environmental pollution, impacts on human health and a decrease in the comfort of living [36]. It is especially important to consider NIE in the cadastral valuation of land, which is the basis for land taxation, since its results shall be socially fair. In the framework of previous studies by Bykova, E.N., methods of NIE assessment in different conditions of the land relations activity in Russia have been developed, which are applicable for both mass valuation and individual appraisal. The relevance of the study is related to the need to improve the methods of cadastral valuation, taking into account the impact of encumbrances to obtain a fair value of land, and to automate this technique to reduce labor and time costs for such an assessment, which corresponds to modern global trends in the development of land valuation work. Therefore, the main goal of this study is to develop a mechanism for automating the process of NIE assessment in the conditions of a non-existent and weak market activity of land relations. In accordance with the goal, the following tasks were set in the study: first, to develop the architecture of the web application “Assessment of Negative Infrastructural Externalities to Determine the Land Value” using the “thin client” technology; second, to design the modular structure of the specified web-application to implement its business logic; third, to develop a conceptual model of the database as the main content component of the projected information system, which makes it possible to automate the process of NIE evaluation in the conditions of a non-existent and weak market activity of land relations.

The scientific novelty of the work is the development of the architectural system of the client-server application, including the development of the fundamental organization of this system; the relationship of elements with each other and with the environment; the principles that guide its design and evolution; and the justification for the use of hardware and software architectural elements for the implementation of the application.

The practical significance of the study is the creation of a web application for calculating land cost, taking into account the encumbrances—“Assessment of negative infrastructural externalities to determine the land cost”—and a heat map of the cadastral value of land in St. Petersburg, taking into account the NIE.

2. Materials and Methods

The methodology of NIE assessment is presented in a previously published study by Bykova, E.N. [36], which involves the use of several methods for the mass valuation of land resources in different conditions of the market activity of land relations: in the absence of market activity—an expert-analytical approach; in conditions of weak market activity—taking into account the degree of coverage of the land plot by ZSTUC, the ratio of the market values of the encumbered ZSTUC lands and those that are free from such encumbrances, obtained by comparing their market sale prices or qualimetric modeling; in conditions of high market activity—modeling by introducing into the model the factor “presence of ZSTUC” based on the grouping of zones according to similar regulations for the use of the territory or by introducing the parameters of this factor.

The automation of methods for the assessment of Negative Infrastructural Externalities is implemented on the basis of developing a “thin client” model for the first two types of market activity in land relations in Russia. The creation of the information system “Assessment of Negative Infrastructural Externalities to Determine the Land Value” included the following main stages: design of the application architecture; software concept development; formation of the database structure.

Designing the application architecture, which is the fundamental organization of the system embodied in its elements, their relationship with each other and with the environment and the principles that guide its design and evolution, includes the stages presented in Figure 1.

The system must be designed based on the criteria of efficiency, flexibility, extensibility, scalability of the development process, testability, reusability and maintainability. To achieve the necessary characteristics of the developed application, at the first stage, a modular structure was created, which allows the product to be presented as a coherent system with uniquely specified result parameters. Decomposition, as a basis, is:

(1)

Hierarchical: a complex system consists of a small number of simpler subsystems, each of which is built from smaller parts, until the smallest parts are simple enough to be directly understood and created,

(2)

Functional: each module should be responsible for solving a specific subtask and performing the function corresponding to it,

(3)

High Cohesion + Low Coupling: implementation of the decomposition quality criterion—focusing modules on solving specific problems with the elimination of dependencies.

The second design phase involves decoupling between modules by creating interfaces and inverting dependencies. At this stage, it is necessary to adhere to two principles: upper-level modules should not depend on lower-level modules, and abstractions should not depend on details; rather, details should depend on abstractions.

The following steps allow for the reduction pf the connectivity of the modules. At the third stage, direct dependencies are replaced by messaging. Sometimes, a module just needs to notify others that some events/changes have occurred in it, and it does not matter what happens to this information later. In this case, the modules do not need to “know about each other”, that is, contain direct links and interact directly, but it is enough just to exchange messages or events.

Next, it is necessary to replace direct dependencies with synchronization through a common core, since there are a large number of modules in the system. Their direct interaction with each other becomes too complex, so the “all with all” interaction should be replaced with a “one with all” interaction.

The last step in the application development is to replace module inheritance with composition, since inheritance has one of the strongest relationships between objects. Therefore, if possible, it should be avoided and replaced with a composition.

The database design includes the steps shown in Figure 2.

There are seven normal forms for relationships between entities, but three are enough for design:

(1)

A relation is in the first normal form if all its attributes are simple; all domains must contain only scalar values. There should be no repetition of rows in the table;

(2)

A relation is in the second normal form if it is in the first form and every non-key attribute depends on the Primary Key;

(3)

A relation is in the third normal form when every non-trivial and left-irreducible functional dependency has a potential key as a determinant.

Among the common types of architectures for the implementation of the author’s information system, the architecture of client-server applications has been chosen, since, in the future, it will be necessary to provide access for potential consumers (clients, appraisers) to the computing powers of the machine that are responsible for complex calculations. This template consists of two parts: a server and many clients. The server component provides services to the client components, while the clients request services from the server which provides those services to the clients. Moreover, the server operates in a waiting mode for client requests. Of the two types of client–server applications (“thin client” technology and “thick client” technology), the first one was chosen, which is justified by the following:

• “Thick clients” work with information based on their own hardware and software capabilities, while thin clients use the central server software only to process data, providing the system only with the required graphical interface for the user work. This means that, sometimes, we can see outdated or not very productive PCs in the role of thin clients. A thin client in computer technology is a computer or client program in networks with a client-server or terminal architecture that transfers all or most of the information processing tasks to the server. An example of a thin client is a browser-based computer used to run web applications.
• A “thick client” is a client that performs manipulations requested by the user independently of the host server. The main server in this variation of the system architecture can be used as a special storage of information, the processing and final provision of which is simply transferred to the user’s local machine. An example is PCs that operate on the basis of their OS and are filled with a complete set of software for the required user tasks [37].
• The “thick client” technology has a number of drawbacks in the implementation of the calculation methodology of NIE assessment, including the economic inefficiency of the development of powerful user systems that are accompanied by energy costs; providing support for complex devices, etc. has to provide a high speed of information processing; the main system has to be in a standby mode while heavy calculations are performed on the client’s side [38];
• In the technology of a “thin client”, computational processes occur on the side of the main server, presenting the client with an interface for interacting with the system [39]. It is the main prerequisite for the use of this particular approach to the design process.

The general architecture of the web application “Assessment of Negative Infrastructural Externalities to Determine the Land Value” is presented by highlighting its main components in Figure 3.

For the implementation of the author’s web application, NGINX was defined as the web server, since its operation speed exceeds the capabilities of Apache by 2–2.5 times, which is confirmed by two performance tests (1000 and 512 simultaneous connections) [37]; the RAM consumption parameters are less than those of Apache [39]; the performance of dynamic content, according to the test results, is the same for the Apache and NGINX servers. It is feasible to use it primarily for static websites and as a reverse proxy for dynamic sites. WSGI was used to interact with the web server with web applications, which justified the use of the Python programming language during the development of the service.

The WSGI server Gunicorn was chosen as the most lightweight and tightly integrated server with the Nginx environment to implement the web application. Gunicornis is a standalone web server with broad functionality; it is based on a preliminary working model, which means that there is a central master process that manages a set of workflows, while all requests and responses are handled entirely by these workflows.

To eliminate any errors in the process, the back-end platform is used, on which the main logic of the web application is applied. The implementation of the back-end component predetermined the synchronous type of frameworks since the information system provides for solving business problems. Django stands out among the main synchronous frameworks in the Python programming language (Django, Pyramid, TurboGears, Web2py, Flask, Bottle and CherryPy), as it has several advantages over the others:

−

Most of the tools for building an application are part of the framework rather than being delivered as separate libraries, while Django contains a huge number of functions for solving the most of web development tasks. For example, the developed application requires ORM, Database Migrations, User Authentication and Administrator Panel;

−

Django as a framework defines a project structure that helps developers understand where and how to add new application logic;

−

Django’s internal implementation has standard security tools and includes mechanisms to prevent common attacks (SQL injection (XSS) and cross-site request forgery (CSRF)).

To visually display the work with business logic, a web application needs a front-end module, which is implemented by the JavaScript framework React.js. It has advantages such as an ease of learning and using, thanks to its simple design; the use of JSX (HTML-like syntax) for templates and very detailed documentation; speed, thanks to the implementation of React Virtual DOM and various rendering optimizations; server-side rendering support, making it a powerful platform for this content-oriented application; support for Progressive Web Apps (PWA), thanks to the “create-react-app” application generator; an ease of testing and the reusability of the code; and the ability to create applications using TypeScript or Facebook’s Flow, which have built-in JSX support.

For use in the project, a relational PostgreSQL database was chosen because the data with which the work will be carried out are structured, while the structure is not subject to frequent changes. In addition, PostgreSQL is free to use, has a wide range of tools, supports extensions for working with spatial data and has a high volume of processed information and ready-made solutions for various tasks.

The support of the spatial data processing on land plots and ZSTUC in the author’s web application is implemented using the open-source software PostGIS. Docker was chosen as a virtualization tool to ensure the inertness of the development environment, which made it possible to achieve the greater performance of the designed system.

The FTP file transfer protocol has implemented the ability to authenticate clients by transmitting a username and password in clear text, to manage data flow and to correct errors, to provide the ability to navigate the directory structure and find files, to create new directory structures or delete directories and, also, to provide the compression of transmitted data.

The initial data for testing the information system are: the results of an interview of experts, obtained personally by the authors of the study; the prices of transactions for the sale and purchase of land plots from the AIS “Monitoring of the Real Estate Market” of Rosreestr [40]; the spatial characteristics of infrastructure facilities and other regime-forming facilities, as well as their ZSTUC from the RGIS of St. Petersburg [41]; and the archive of St. Petersburg SBI “City Department of Cadastral Valuation” reports of the previous round of cadastral valuation in St. Petersburg [42].

The finished application implements a role-based approach to real estate valuation. Upon request, the authors can provide access to the system. The order of interaction between the administrator and the system is shown in Figure 4.

The application allows the expert to perform a series of simple steps, as shown in Figure 5.

3. Results

3.1. The Logic of the Web Application “Assessment of Negative Infrastructural Externalities to Determine the Land Value”

To implement the business logic of the web application, a modular structure was designed, where each part of this structure is responsible for one global task. Conceptually, the following groups of modules are highlighted:

• Administration module providing management functions (creating and editing users and their roles, differentiating access rights to data and system functions, etc.);
• Module for processing the geodata, including all spatial data of the public cadastral map;
• Module for processing the market data on land plot transactions;
• Module for processing the data of the expert’s questionnaire to determine the regulation coefficient in the ZSTUC;
• Calculation module that implements the methodology for calculating the cadastral value in various conditions of market activity;
• Module for working with database models, which works with a database using ORM.

The diagram of the modules is shown in Figure 6.

The administration module AdminPanel is intended for data management, providing operations for creating and editing users and their roles. For example, Figure 7 presents an interface with a function that creates a record about an expert for the implementation of an expert-analytical approach to the appraisal of NIE. This module is responsible for obtaining calculated data based on the results of processing all related information, providing them in xlsx format.

The geodata processing module LinkerGeodata consists of several submodules: GeometryHandler, CostAttributeHandler and EncumbranceAttributeHandler. Each of the submodules is responsible for its functions: the GeometryHandler submodule is responsible for processing the geometry of entities, the EncumbranceAttributeHandler submodule is responsible for processing the attribute data associated with encumbrance entities and the CostAttributeHandler submodule is responsible for obtaining cost data from the attributes of the geoinformation layer. The functions of each of them are presented in

Table 1. Characteristics of the submodules of the LinkerGeodata geodata processing module.

Submodule Element	Submodule Element Functionality
Submodule “GeometryHandler”
ParcelEntities	is responsible for processing the geometry of land plots, using various geospatial functions for calculating the intersection of areas, spatial entries, intersections, etc.
	is responsible for the functionality associated with the processing of the attributive information of objects of the geoinformation layer
	allows you to write the received data into the existing database model
ReceiverEstimateFactors	is responsible for processing entities that reflect source objects by pricing factors
	assigns the values of pricing factors to land plots
EncumbranceEntities	is responsible for processing the geometry of the ZSTUC
	similar in functionality to the ParcelEntities module
Submodule “EncumbranceAttributeHandler”
TypeReceiver	is responsible for the connection of ZSTUC entities with user data, while indicating the type and weight
AncillaryDataHandler	implements the preparation of auxiliary data on ZSTUC for further calculation
	the original records in the table are grouped to give each type of ZSTUC a certain weight
Submodule “CostAttributeHandler”
-	is responsible for obtaining value data from the attributes of the geoinformation layer
	the module implements functions for filtering and converting cadastral value data into the format required for further calculations

.

The MarketDataHandler module for processing market data includes the CorrectionIntroducer and StatisticalAnalyzer submodules (

Table 2. Characteristics of the submodules of the MarketDataHandlermodule.

Submodule	Submodule Functionality
CorrectionIntroducer	introduction of various adjustments to market data
StatisticalAnalyzer	checks for the representativeness of the sample

).

The questionnaire data processing module “Questionnaire” consists of several submodules: WeightReceiver, FocusGroupHandler and FormDataHandler. The data come from the form of the web application, which the experts fill out (an example of the interface is shown in Figure 8 and Figure 9).

It should be emphasized that the methodology for taking into account NIE, arising from the presence of encumbrances on land, provides for taking into account not ZSTUC but prohibitions and restrictions that determine the rules for the land use in them (zones). Therefore, the web application uses prohibitions and restrictions, which are obtained on the basis of extraction from legislative acts for each ZSTUC (protection zones of main pipelines, power lines, railways, roadside lanes of highways, sanitary protection zones of industrial enterprises, water protection zones, coastal protective strips, etc.) and a comparison with the type of permitted land plot use of a certain market segment. For garden lands, such prohibitions and restrictions are presented in Figure 9.

The functions of each of the submodules of the Questionnaire module are presented in

Table 3. Characteristics of the submodules of the Questionnaire module.

Submodule	Submodule Functionality
WeightReceiver	converts matrix data into a representation of weights required for subsequent calculations
FocusGroupHandler	designed to manage the qualifications of members of the expert group allows you to assess the competence of an expert, followed by making decisions on the data provided
FormDataHandler	designed to convert data provided by experts into a format for presenting the information in a matrix form

:

To carry out all calculations as part of the valuation, the CostEstimator calculation module was developed, the submodules of which implement the following functions (

Table 4. Characteristics of the submodules of the CostEstimator module.

Submodule	Submodule Functionality
DepressiveMarketEstimator	implements the methodology for calculating the cadastral value in the absence of market activity
InactiveMarketEstimator	implements the methodology for calculating the cadastral value in conditions of weak market activity

):

To implement the functions of the DepressiveMarketEstimator and InactiveMarketEstimator modules indicated in Table 4, the techniques presented in the monograph [43] were used.

The methodology of NIE assessment in the absence of market activity provides for stages 1–8, the implementation of which is not automated. At the first stage, the permitted types of activities are determined according to the intended purpose of the land:

• for lands of settlements, the main and conditional types of permitted use are in accordance with the Rules of Land Use and Development (RLUD);
• for lands of other categories, they are in accordance with the classifier of the types of permitted use of land plots and the Unified State Register of Real Estate.

The definition of significant ZSTUCs for lands of the considered type of use includes:

• Assessment of the impact of the ZSTUC regulations on permitted activities by comparing them to each other;
• Estimation of the frequency of the occurrence of ZSTUC on land plots of a certain type of use $R_i(Q)$ based on spatial analysis of the territory and on the determination of statistical probability $P_i(Q)$:

  $$P_i(Q)=\frac{g_{ZSTUCi}}{G_{LP}}$$

  (1)

  where G_LP is the total area of land plots; g_ZSCUTi is the total area of the i-th ZSTUC as a measure expressing the number of favorable event Q outcomes;

  $$R_i(Q)=\frac{g_{ZSTUCi}}{G_{ZSTUC}}$$

  (2)

  where R_i(Q) is the relative frequency of the occurrence of a certain type of ZSTUC on land plots; G_ZSTUC is is the total area of all ZSTUCs.

Based on the study of the regulatory acts of the Russian Federation on ZSTUC, systematization and analysis of the regulations of ZSTUC are carried out, including the identification and structuring of all types of ZSTUCs, the grouping and codification of prohibitions and restrictions in ZSTUCs and the analysis of the regulations of each ZSTUC for the recurrence of prohibitions and restrictions, synonymy, essential unity, etc. Further, the ZSTUC regulations are compared with possible types of activities on the land plots of the considered type of use.

Expert studies begin with the formulation of the purpose of the analysis, which is to study the impact of the existence of ZSTUCs and the prohibitions and restrictions in them on the efficiency and completeness of land use. To use an expert approach to NIE assessment, the important stages are the construction of a high-quality hierarchical model, the development of an expert questionnaire and the selection of experts.

The hierarchical model includes three levels: goal (fifth stage of the methodology); level II factors—grouping of prohibitions and restrictions by types of agricultural activities based on three stages of the methodology; level III factors—restrictions and prohibitions of agricultural activities in the ZSTUC (stages 3–4). The advantage of this structure is the possibility to calculate the coefficients of the influence of specific prohibitions and restrictions.

The expert questionnaire was drawn up according to a hierarchical principle and contains 10 tables, in the first of which there is a pairwise comparison of the groups of prohibitions and restrictions. In the other tables, the prohibitions and restrictions are compared in the context of each group of the type of economic activity. The form of the developed expert questionnaire is presented in Appendix A of this article [43].

The selection of the experts’ composition for the appraisal is the most critical stage of the methodology, since the experts shall be competent to fulfil the purpose of the study. The general requirements for experts have been determined: competence in their field of research; they should not make a decision based on the information received; direct involvement in the land use sphere; combination of narrow specialization with a broad outlook; availability of production or life experience. Depending on the type of permitted use of land, specific requirements are imposed on the composition of experts. For example, for agricultural land, those specialists are selected who are directly working in agriculture or are managing these lands (employees of agricultural enterprises, heads and members of peasant (farmer) households); for horticultural lands, the above-mentioned experts have been supplemented by the current chairmen and members of horticultural and horticultural non-profit associations and the title holders of such plots. In the latter case, a requirement has been added to the term of ownership of rights to land plots (at least 5 years).

The automated steps of the methodology are presented in the form of

Table 5. Automated stages of the methodology for NIE assessment (in the absence of market activity).

Stages	Description
9. Obtaining the expert assessments	Pairwise comparison of factors in relation to their impact (weight or intensity) on the efficiency and completeness of land use (nine-point scale of relative importance). When filling out the matrices, the degree of influence is determined: 1. Groups of factors of the II level for the purpose of the analysis (I level) with the answer to the question: “Which group of prohibitions on activities has the greatest impact on the efficiency and completeness of land use?” 2. Bans for each activity in the bans group by answering the question: “Which type of prohibition on activities has the greatest impact on each type of activity?”
10. Processing the information received and determining the degree of consistency of the expert opinions	1. Estimation of the components of the eigenvector of each factor: $a_i=\sqrt[n]{{\displaystyle \prod _{j=1}^n{a}_{ij}}}=\sqrt[n]{a_{i1}\cdot \dots \cdot a_{ij}\cdot \dots \cdot a_{in}}$, where aij is elements of the matrix of answers of experts A, and n is the number of compared elements (influencing factors). 2. Normalization of expert evaluations: $w_i=\frac{a_i}{{\displaystyle \sum _{i=1}^n{a}_i}}$. 3. Calculation of the sums of elements аij by columns: $b_{ij}=a_{1j}+\dots +a_{nj}={\displaystyle \sum _{i=1}^n{a}_{ij}}$. 4. Calculation of the maximum eigenvalue of the matrix (λmax): ${\lambda }_i={\displaystyle \sum _{i=1}^n{b}_{ij}\cdot w_i}$$; {\lambda }_{\max }={\displaystyle \sum _{i=1}^n{\lambda }_i}$. 5. Calculation of the Index of Consistency (IC): $IC=\frac{{\lambda }_{\max }-n}{n-1}$. 6. Determination of the Average Consistency (AC) obtained by the random selection of quantitative judgments from the scale 1/9,1/8,..., 1, 2,..., 9, but with the formation of an inversely symmetric matrix (Saaty, 1993). 7. Calculation of the Relationship of Consistency $ROC=\frac{IC}{AC}$ and assessment of the quality of the expert.
11. Calculation of the weighting factors kbv of prohibitions and restrictions on activities in the ZSTUC	Determination of kbv prohibitions and restrictions of activities: $k_{\beta \nu }=W_i\cdot w_i$ where Wi is the weight of the group of prohibi-tions and restrictions; wi is the value of the con-tribution to the group of the prohibitions and re-strictions themselves.is the value of the contribution to the group of the prohibitions and restrictions themselves. Analysis of the spread and consistency of expert estimates: ▪ variation range: $R=x_{\max }-x_{\min }$ where xmax is the maximum estimate of the factor; xmin is the minimum estimate of the factor; ▪ root-mean-square deviation (RMSD): $\sigma =\sqrt{\frac{{\displaystyle \sum _{j=1}^m{(x_j-{\overline{x}}_{cp})}^2}}{m-1}}$ where xj is the estimate given by the j-th ex-pert; m is the number of experts;${\overline{x}}_{av}$ is the average value in the row; ▪ coefficient of variation: $V=\frac{\sigma }{{\overline{x}}_{av}}\cdot 100\%.$ Experts’ opinions are agreed upon at V < 33%.
12. Calculation of regulation coefficients Ki for each ZSTUC and modifications of their overlapping at intersections	Regulation coefficient showing the remaining efficiency of using the encumbered part of the land plot: $K_i=1-{{\displaystyle \sum _{i=1}^n{k}_{\beta \nu }}}_i$, where n is the number of prohibitions and restrictions.

.

The methodology for NIE assessment in conditions of weak market activity is fully automated in the InactiveMarketEstimator submodule and includes:

Processing of the initial cartographic base and initial market information; the formation of data samples with the presence and absence of ZSTUC;

Identification of pricing factors of the market value of land plots;

Correction of market data by comparison elements for the terms of sale (discount on the offer, on the date of sale, on the transferred property rights, on the terms of sale);

Ranking the values of pricing factors and calculating the relative quality indicators:

$$Q_i=\frac{l_i-q_r}{q_{ref}-q_r}$$

(3)

where Q_i is the relative indicator of the quality of the i-th analog object (land plot); l_i is the absolute or rank value of the cost factor of the i-th analog object; q_r is the rejection (minimum) value among the values of one factor for analog objects; q_ref is the reference (maximum) value among the values of one factor for analog objects;

Determination of weights and weighted relative quality indicators of pricing factors:

To calculate the weights (G_i), the condition of maximization of the determination coefficient is used, which shows the quality of the cost calculation model depending on the quality coefficient, when the array of weight coefficients changes, subject to their sum being 100%.

The weighted relative quality scores are determined by multiplying the weights obtained by the Q_i values.

Checking pricing factors for multicollinearity (determining the partial correlation coefficients (PCC) between factors and the cost of land, as well as the PCC between the factors themselves).

In the presence of a multicollinearity of factors (PCC > 0.75), the one that is least correlated with the market value is excluded, and the procedure for determining the weights of pricing factors is repeated.

Determination of Integral Quality Factors (IQF) for each land plot:

$$IQ{F}_i={\displaystyle \sum _{i=1}^p{Q}_i}\cdot {\displaystyle \sum _{i=1}^p{G}_i}$$

(4)

where G_i is the weighting factor of the pricing factor.

Construction of land value models with and without taking into account the presence of a certain type of ZSTUC (regression analysis);

The calculation of the regulation coefficient is carried out according to the formula:

$$K_i=\frac{Y_{with\hspace{0.17em}ZSTUC}}{Y_{without\hspace{0.17em}ZSTUC}}$$

(5)

where Y_{with_ZSTUC} is a specific indicator of the cadastral value of a reference land plot, taking into account ZSTUC; Y_{without_ZSCUT} is a specific indicator of the cadastral value of a reference land plot without accounting for ZSTUC.

The last stage of the methods is the calculation of the Cadastral Value (CV) of land, taking into account the regulation coefficient (prohibitions and restrictions on economic activity), which is carried out according to the formula:

$$CV={\displaystyle \sum _{i=1}^n(SICV\cdot S_{ZSTUCi}\cdot K_i)+SICV\cdot S_{unenc}}$$

(6)

where SICV is a Specific Indicator of the Cadastral Value by the market segment; S_unenc is the area of the unencumbered part of the land plot; n is the number of encumbered parts of the land plot formed by the ZSTUC, including their imposition at the intersection.

The Models module is responsible for working with the database using Object-Relational Mapping (ORM), the purpose of which is to transform the description of the data structure and work with them from an imperative programming language to a declarative one.

Thus, the aggregate work of the modules allows for the implementation of the methodology of calculating the cadastral value of land plots in conditions of weak market activity and graphically displaying the assessment objects and the results of calculating the values in the form of a heat map with extremes (Figure 10 and Figure 11, Video 1).

3.2. Database Structure of the Information System “Assessment of Negative Infrastructural Externalities to Determine the Land Value”

UML applications are used to display the structure of the database. Entity structures have been developed for each group of models to provide information support for the processes.

The group of models of the questionnaire process includes the entities shown in Figure 12. The SubFactor and BaseFactor entities are a repository of information on the factors being compared. To implement the methodology, the essence of the factor store is divided into two components, reflecting the main and secondary groups of factors, respectively. The entities ExpertAnswerSub and ExpertAnswerBase are data stores for expert responses and serve the purposes of implementing the calculation methodology and reflecting the retrospective.

The user model group includes the entities shown in Figure 13. The User entity stores information about existing users with a brief description of them: UserAuth—information about user authentication, where the hash function login modifier is specified; StatusAccount—information about the existing types of user roles (required to restrict access to data); and PassSurveyInfo stores information about passing the survey users.

The group of calculated data models includes the entities ExpertData, ResultData, Parcels and MarketData, which is justified by the need for calculations and storage of information according to methods, according to [43]. The specified structure is shown in Figure 14.

As can be seen from Figure 8, the ExpertData entity stores the results of processing the experts’ data, the array of experts who provided the data and the time of the examination; the ResultData entity stores the cadastral values obtained using different methodologies, as well as the data on the basis of which the calculations were made. Information about the geometry of land parcels with accompanying attribute data is contained in the Parcels entity, while the information about market data (transaction type, offer price, transaction terms, transaction date, geometry) is contained in the MarketData entity.

Groups of models by encumbrances (entities Encumbrance and TypeEncumbrace) and by source objects of pricing factors (entities FactorSourceEntities and TypeFactorSource) are similarly created (

Table 6. Characterization of the entities of the groups of models by encumbrances and source objects of pricing factors.

The Essence	Information
Model group on encumbrances
Encumbrance	Information about the geometry of encumbrances with accompanying attributive information
TypeEncumbrace	Information on the names of the types of encumbrances and the values of the weights corresponding to each type of encumbrance
Group of models by source objects of pricing factors
FactorSourceEntities	Information about the entities of source objects of pricing factors (type of source object, geometry, factor values)
TypeFactorSource	Information about the type of source object of pricing factors with a brief description

).

The aggregate work of the modules is presented on Google Drive: https://drive.google.com/file/d/1w4DJPIIRE35bmz8dk-k9bVB5pNWvs3cN/view?usp=sharing (accessed on 1 July 2022).

The program code of the web application was implemented using the Python programming language. A fragment of the program code is shown in Figure 15.

As a result of the research, the following scientific and practical results were obtained:

• The need to automate the NIE assessment in the context of various activities of the real estate market to determine the value of land has been substantiated. The use of modern approaches to automation, as well as the use of spatial analysis of the territory based on GIS technologies, computerized databases and programming languages, will increase the efficiency of the assessment work and ensure a higher accuracy of results by reducing the influence of the human factor and formalizing certain assessment algorithms.
• The architecture of the web application “Assessment of Negative Infrastructural Externalities to Determine the Land Value” based on the “thin client” technology was developed and substantiated. The proposed architecture contains a web server, a WSGI server, a back-end platform, a front-end module, an FTP server, a PostgreSQL relational database, PostGIS and Docker software, and it allows you to fundamentally organize the interaction of all structural elements of the designed information system.
• The modular structure of the specified web application is implemented, allowing for the implementation of its business logic. The proposed set of modules includes the functionality that is necessary for NIE assessment in order to determine the value of land and provides the management of information system users and their roles, the processing geodata, the market data on land transactions, the expert questionnaire data, the work with a data bank as well as the implementation of the methodology for calculating the cadastral value in the conditions of a non-existent and weak market activity of land relations.
• The conceptual model of the database has been developed, which, in the future, will become the information core of the system being designed. The proposed model includes several groups (a group of questionnaire process models, a group of user models, a group of calculated data models, a group of models by encumbrances and by source objects of pricing factors) and allows for automated NIE assessment in determining the value of land.
• The practical implementation of the architecture of the web application, its modular structure and the conceptual database model was carried out using the Python programming language.

4. Discussion

In the modern conditions of the need for land assessment work digitalization and automation, the development of new ways to obtain accurate assessment results in a short period of time takes on a special role for the regulation of land relations.

A comparison of the results of the author’s software solution presented in this paper with similar developments made it possible to note:

Firstly, in the mobile GIS application by Aydinoglu, A.C. and Bovkir, R. [44] designed for various thematic analyses of information about the earth, there is a similar architecture of a thin client. The difference is that, instead of a web interface, a mobile application is used. At the same time, the presented author’s development has the advantage of the ability to edit attributive information and the function of expert group questioning, but it does not provide for the use of fuzzy sets to calculate pricing factors.

Secondly, the structure of a deep learning artificial neural network for spatial modeling in the process of mass evaluation, used by Wei, C., Fu, M., Wang, L., Tang, F. and Xiong, Y. [45], allows for the consideration of big data and has similarity in creating a database and applying spatial analysis with the presented development, but it differs in the methods of creating the architecture of a GIS application.

Thirdly, this development, unlike the web application for automating real estate valuation by Op’t Veld, D., Bijlsma, E. and Van De Hoef, P. [46], allows, in addition to automating the process of mass valuation, one to take NIE into account.

The advantage of these studies is the algorithm for taking into account price inflation for real estate when assessing it. The disadvantage is the lack of entering your own market data. The project uses exchange interfaces with real estate agencies and specialized agencies, which can potentially cause problems with the quality of the information used. Automated plausibility check algorithms are used as its control.

The results of the research correlate with the findings of Clark, S.D. and Lomax, N. [47], who suggest that the use of web applications and machine learning for data acquisition, processing and calculations can improve the efficiency of real estate valuation and labor productivity.

As can be seen from the above, the developed algorithm is a promising direction for the practice and theory of land valuation work, combining the basics of the technology for taking into account negative infrastructural externalities in the assessment and automation of the assessment process itself. The advantages of this web application are the wide range of land restrictions taken into account and the automatic calculation of the land cost, taking into account the opinions of experts.

The main advantages of the author’s concept of the architectural system for automated land assessment are:

First, the flexibility and versatility of the expert approach implementation. This makes it possible to assess the land cost, taking into account the NIE, as well as by changing the settings (the purpose of the assessment, factors, objects of study, etc.) to obtain the results of an examination for the purposes of forecasting and planning the development of territories, making managerial decisions in the field of land use;

Second, the convenience and simplicity of the visualization and interpretation of the survey of experts in the created environment of the software product;

Third, the unity and structure of the database on the assessment objects and expert results with the possibility of spatial visualization in GIS;

Fourth, a single automated workspace for both the appraiser and the participants involved in the assessment activity;

Fifth, a cross-platform of the web application, allowing for the use of any operating system, including open source systems, which reduces software costs;

Sixth, the implementation of a thin client, which allows for a reduction in the cost of implementing the application in production; because the application hardware is not demanding in terms of computing power, heavy operations are delegated to the server.

Seventh, the implementation of the concept of experts’ intuitive interactions with the questioning process in the form of a web application module allows for a reduction in the time spent on questioning.

The disadvantages include: the need for qualified personnel who are able to expand the system to suit the needs of appraisers; the spatial data editing interface has a number of difficulties, so the preparation of geoinformation layers is implemented in other software; the lack of integration with tools familiar to the appraiser, such as MS Excel; the use of a relational database to store information, which complicates the use of cloud computing. In addition, evaluation activities are often accompanied by unstructured information, which causes the problem of its processing when choosing a relational database type.

As promising areas of research, it is envisaged to introduce an information system for the purposes of cadastral and individual land valuation to improve the interface of the web application in order to meet the requests of potential users of the proposed product (appraisers, real estate agents, land plot title holders), as well as to expand the tasks of NIE assessment in an active land market.

According to the results of testing, the application will reduce the labor and time costs of the appraiser by 66% of the total cycle of work by automating the stages of developing expert questionnaires (partially), delivering questionnaires to experts, processing questioning results, calculating the coefficients of prohibitions and restrictions on the activity of the ZSTUC, regulating coefficients showing the remaining efficiency of using the encumbered part of the land plot, performing spatial operations to isolate the necessary attributes and calculating the cadastral value of the land plot.

The value of this research is the perspective opportunity to use its results for a wide range of subjects in the field of real estate valuation, including state budgetary institutions of Russian Federation subjects in the process of cadastral valuation and private valuation organizations in the individual valuation. Elements of the research have been introduced into the educational process of St. Petersburg Mining University in the preparation of baccalaureates—21.03.02 “Land management and cadasters” direction (preparation profile “Urban cadaster”)—and masters of direction—21.04.02 “Land management and cadasters” direction (preparation profile—“Real estate management and integrated development of the territory”). Such academic creativity contributes to the greater integration of the competency-based approach to learning with modern realities and the requirements of production activities.

5. Patents

In the Patent Office of Russia, the authors of this article received an author’s certificate for a computer program for NIE assessment in the context of an inactive market turnover of land resources. Certificate number RU 2022611081 [program for calculating the historical and cultural value of urbanized territories coefficient. Available online: https://elibrary.ru/item.asp?id=47785421 (accessed on 1 July 2022)].