ERSDMM: A Standard Digitalization Modeling Method for Emergency Response Based on Knowledge Graph

Abstract

Standard digitalization is a crucial step in social and economic development and the transformation of digital technology. Standard digitalization is of great significance in the promotion of sustainable economic and social development. This paper proposes a standard digitalization modeling method for emergency response (ERSDMM) based on knowledge graph (KG). Firstly, this paper analyzes the knowledge structure of emergency response standards (ERS) and constructs a “seven-dimensional” model of ERS based on the public safety triangle theory. An ontology model of the emergency response domain is then created. Secondly, ERS and emergency scenario fine-grained knowledge are extracted. Thirdly, a standard reorganization model is constructed to meet the needs of the scenario response. Finally, the ERSDMM is applied to the GB 21734-2008, which proves that the ERSDMM is available. Taking RES as an example, this paper explores the path and practice of standard digitalization. ERSDMM solves standards-related problems, such as overlapping content, coarse knowledge granularity, incomplete coverage of elements, and difficulty in acquiring knowledge.

1. Introduction

Emergency management is an integral part of enhancing national governance capacity. As a critical component of emergency management, emergency response standards (ERS) are a solid guarantee and essential support for improving the governance system. ERS can promote the transformation and integration of system construction and governance efficiency. Digital transformation is the development trend of the digital economy era. The concept of “digitalization” is introduced into the field of standardization, and thus “standard digitalization” is formed [1]. Standard digitalization is an inevitable trend for standardization work to further the development of global digitalization [2].

Digital technology has been applied in emergency plans in China. However, digital technology has not been combined with ERS, and the digitalization of ERS has not been achieved. There are some problems in ERS digitalization transformation in China: (1) the process of standard establishment relies on expert knowledge, and standard reading is inefficient. Currently, the updating, revision, and analysis of standard documents often depend on knowledge specialists. The format of most standard documents is PDF. There is ambiguity in standard searches, reading, and comprehension [3]. Standard documents in PDF format impose a reading burden for both the documentation supervisors and users [4]. (2) There are some problems with the content of standards, such as cross inconsistency, untimely updating of standards, incomplete coverage of standardization elements, etc. (3) The granularity of standard knowledge is coarse. The mismatch between document-based and unit-based knowledge becomes obvious [5]. It is not easy for users to obtain knowledge and select decision indicators.

Knowledge graph (KG) has strong applicability in the standardization field, and can resolve problems in applying existing standards. We can use KG to meet the requirements in the process of standard digitalization. KG is the technical basis for the digitalization of ERS. As an effective means of knowledge interconnection, the correlation matching between ERS elements and the emergency response scenario elements can be effectively realized by using KG, which lays a solid foundation for the realization of emergency assistant decision making. This paper proposes a framework for a standard digitalization modeling method for emergency response (ERSDMM). We can use ERSDMM to solve the problems of ERS digitalization and realize the effective use of standard knowledge. Through this framework, we can obtain knowledge of ERS and emergency scenarios, and realize standard knowledge reorganization based on emergency response. ERS knowledge includes two granularity levels of knowledge. Standard structure knowledge is a common element of standard composition structure and is the embodiment of coarse-grained knowledge. Standard content knowledge is fine-grained knowledge of standard paragraph content. Using ERSDMM, we can realize the digitalization of ERS to obtain fine-grained knowledge, which can meet various needs. The implementation of standard digitization can improve the standardization and accuracy of emergency responses, which is of great significance to improving the emergency response capacity of China.

2. Development Status

2.1. The Trend of Standard Digitalization

Standard digitalization refers to the application of digital technologies to the standard and life cycle of standardization work, and the creation of a new machine-readable standard form. As a result, the digitalization and intelligence of the standard is realized [6]. Standard digitalization started as a standard full-text search [7,8]. Currently, it refers to converting standard documents into a machine-readable, usable, and resolvable form [9] and aims at human reading [10]. It pays more attention to the semantic connections between entities in standard documents [11]. The International Electrotechnical Commission (IEC) and the International Organization for Standardization (ISO) classify machine-readable standards into five levels: traditional text formats (such as paper documents), open digital formats (such as PDF), machine-readable files (such as XML), machine-operable content, and machine-resolvable content. Many international standard organizations use this model to develop standards for digital implementation.

Standard digitalization can realize the mapping and fusion of standard knowledge, which plays a vital role in responding to users’ needs. This has become a hot topic and development direction of ISO, the European Committee for Standardization (CEN), the European Committee for Electrotechnical Standardization (CENELEC), and some countries. ISO is the largest and most authoritative international standardization organization in the world. ISO proposes the SMART standard strategy and puts forward the overall idea for the formulation and implementation of future standards. In 2021, ISO issued the ISO Strategy 2030, proposing that digital technology is one of the driving factors of ISO reform, and the way to create, format, and deliver ISO standards should change. CEN and CENELEC actively promote the SMART standard strategy. CEN and CENELEC were studied earlier than ISO. In 2017, they launched the CEN-CENELEC Digital Transformation Strategic Plan, which proposes to ensure that the standardization needs of digital transformation in industry are met through standard digitalization. In 2018, they launched three projects: online standardization, future standards, and open-source innovation. In 2021, CEN and CENELEC issued the CEN-CENELEC Strategy 2030. Strategy 2030 introduces standards such as requirements for machine-applicable, readable, and transferable content, and promotes fine granularity in standard content.

The United States, Britain, Germany, and China are also actively promoting standard digitalization. The American National Standards Institute (ANSI) officially issued the United States Standards Strategy (USSS 2020), proposing that digital technologies can be effectively used to optimize the formulation of global standards. The British Standards Institute (BSI) proposed a rule-based online standard formulation method in 2020. The standards formulated through the standard formulation method can be called BSI Flex standards. The development of standard digitalization in Germany is inseparable from the industry 4.0 strategy. At this stage, Germany is gradually shifting the focus from standard digitalization to the digital twin field. After long-term use of standardization practice, China has realized the significance of standard digitalization. The Chinese National Standard Library has developed a comparison system for Chinese and foreign standards. The China Electronics Standardization Institute proposed and initiated the IEEE standard P2959, called Standards for Technical Requirements of Standard-Oriented Knowledge Graphs. P2959 emphasizes the application direction of knowledge graphs in the process of standard digitalization. In 2021, China issued an outline to promote standardized development at the national level. The document set a goal of continuous improvement of standard digitalization and proposed the development of machine-readable standards and open-source standards.

Currently, the mainstream research direction is from machine-readable files to machine-operable content. It aims to achieve essential machine operation, such as question and answer systems [12,13]. The file structure and formula [3] in the standard have been realized as machine-readable. However, in the field of emergency, there is a shortage of text knowledge representation of ERS content. ERS are existed in the form of electronic text in China’s emergency management information system, which needs to be manually queried during the emergency response procedures. ERS in electronic text form brings inconvenience to emergency management. From the perspective of emergency needs, the proposed ERSDMM can achieve the machine-readable standard and meet users’ query and intelligent decision-making needs.

2.2. Technical Status

Standard digitalization technologies include XML and knowledge graphs. ISO has developed the ISO Standards Tag Set (ISOSTS), which describes the standard full-text content and metadata. ISOSTS provides a common format for publishing and exchanging standard content. Based on ISOSTS, the National Information Standards Organization (NISO) of the United States has enriched and optimized it to form the NISO Standards Tag Set (NISOSTS). In 2019, China published a national standard, XML-based structured labeling framework for national standards (GB/T 37967-2019), which specifies the XML tag set of standard text structure.

A KG is the form of the deepening development of a semantic network. By extracting information from semi-structured or unstructured data to form triples (subject, predicate, object), we can construct KG [14]. A KG involves a large number of entities and their relationship types [15]. Real-world entities and their interrelations can be described and organized in a graph [16]. A KG can process multisource heterogeneous data more quickly and efficiently and perform correlation analysis [17]. A KG, also known as a knowledge base, is a structured representation of facts [18]. There are differences and connections between knowledge graphs and knowledge bases. (1) In essence, KG is a kind of knowledge base called a semantic network [19] and is a knowledge base with a directed graph structure. (2) KG pays more attention to the instance data. It contains semantic information, which can be used for inference, and it has a more flexible form and better scalability. We can use KG to realize semantic organization and association, and mine deep knowledge of different granularity knowledge units [20,21].

For the application of KG in the standard, Lv et al. [22] proposed an open relation extraction model in the agricultural products field based on dependency parsing and constructed a KG in the agricultural product standard domain to provide additional analysis support for the supervision of agrarian product standards. Qin et al. [23] extracted the reference relationship between the content and the document of national food safety standards (NFSS) to construct KG, which is a convenient tool for understanding the relationships in NFSS. Based on NFSS KG, Zhang et al. [24] used the BERT-TCNN-BiLSTM model to extract common elements, such as terms, names and scopes, as well as the relations between release, implementation, and drafting in the standard content. They construct a KG of green standards and strengthen the application. Jiang et al. [25] created a knowledge graph of construction safety standards (CSS) which includes five levels of concepts and eight types of relations, and improves the knowledge management of CSS.

In conclusion, XML tags and KG have been studied in standards. Standard structure knowledge can be obtained through XML tags. Fine-grained knowledge of standard content can be obtained by building a KG. The realization of common element extraction and knowledge reasoning provides the basis for the digitalization of ERS. However, in addition to common elements such as terms and scope, there are also many emergency management elements in ERS. The extraction of emergency management elements needs to be further researched.

3. Construction of ERSDMM

According to the normative and structural features of standard documents, the KG can better show the correlation among standards and resolve problems in the application of standards. As a relatively standardized information resource, the characteristics of standard documents can meet the requirements of KG. Standard KG can be used for reorganization and knowledge mining of standard content [26]. By constructing a standard KG, we can better analyze and research the standard documents [27]. A standard KG can use information-based means to serve standard setters, managers, and users.

Therefore, this paper proposes an ERSDMM. The construction process of ERSDMM consists of three parts, including the construction of an emergency response domain ontology model, knowledge extraction, and knowledge reorganization. The process is shown in Figure 1.

3.1. The Construction of Emergency Response Domain Ontology Model

This paper aims to build a standard digitalization model for emergency response. The construction of emergency response domain ontology is the first step to realizing ERS digitalization. The emergency knowledge includes ERS knowledge and emergency scenario knowledge. Classifying emergency knowledge and constructing an ontology model are essential tasks to realize standard digitalization. How to build an emergency response domain ontology to comprehensively cover fine-grained knowledge is a problem to be solved.

3.1.1. “Seven-Dimensional” Model

Based on the public safety “triangle” model and knowledge system framework of ERS, this paper constructs an “seven-dimensional” emergency response domain model. We comprehensive cover emergency response domain knowledge in seven dimensions.

As an organizational program of emergency management standards, the emergency response standard system describes the structure and level of ERS, and indirectly describes the composition and connection of ERS knowledge. It is a high-level summary of emergency response knowledge. Currently, a unified and complete public safety emergency response standard system does not exist in China. In academia, scholars build emergency response standard system frameworks based on emergency management processes, such as the public safety emergency standard system [28] and the urban community emergency management standard system [29].

The public safety “triangle” model is an essential basis for constructing an emergency response standard system. The three sides represent emergencies, hazard-affected carriers, and emergency management. The effects are called disaster factors [30], as shown in Figure 2.

According to the construction stage of the standard system, the emergency scenario, and the ERS content, the “seven-dimensional” emergency response domain model is established based on the “triangle” model.

In the triangle model, emergency management involves taking measures in response to emergencies. It includes the topics of emergency procedures, measures, and implementation. Therefore, we expand emergency management to the “emergency procedures” dimension, “measures” dimension, and “subject” dimension. The four emergency management processes include mitigation, preparedness, response, and recovery [31], which cover the whole life of an emergency. Because most ERS are formulated according to different procedures, “emergency procedures” is treated as a separate dimension. The “measures” dimension describes the response measures taken in emergency procedures. We can get the measure indicator basis obtained from ERS to support decision making. The implementation of measures requires related subjects, so the “subject” dimension is created. The emergency is the accident itself, and the disaster factors are the elements that cause the emergency. The description of the emergency scenario includes type, time, location, and disaster factors. Therefore, the emergency and disaster factors are combined into the “object” dimension. The hazard-affected carriers are the object of the emergency. The “hazard-affected carriers” dimension is added. Emergency supplies refer to the material support needed in the emergency response procedures, and play a vital role in emergency management. In ERS, the emergency supplies standards form an independent system. The “Emergency supplies” dimension is added. The “Reference” dimension is added to provide the source of indicator knowledge. The emergency response domain “seven-dimensional” model is established, as shown in Figure 3.

3.1.2. Construction of Emergency Response Domain Ontology

The digitalization of ERS is intended to organize the ERS and emergency scenario knowledge with digital technologies to obtain fine-grained knowledge. It can realize knowledge decision reasoning and emergency response. This paper constructs an emergency response domain ontology based on the seven-step method and the seven-dimensional model. We first design the conceptual hierarchy according to the seven-dimensional model. Then, we build the relations between concepts and finally create the emergency response domain ontology.

Concept Design

Based on the “seven-dimensional” model, this paper selects seven primary concepts, including the subject, object, hazard-affected carriers, emergency procedures, measures, emergency supplies, and reference. To ensure that the granularity of the emergency response domain knowledge meets the application requirements, it subdivides the seven primary concepts, as shown in

Table 1. Concept level of the subject.

Primary Concepts	Secondary Concepts	Tertiary Concepts
Subject	Government organization	Government departments, emergency command agencies, etc.
	NGO	Social groups, volunteer teams, etc.
	Unit	Managers, technicians, etc.

,

Table 2. Concept level of the object.

Primary Concepts	Secondary Concepts	Tertiary Concepts	Fourth-Level Concepts	Reference
Object	Emergency	Natural disaster	Flood and drought disasters, meteorological disasters, etc.	GB/T 35561-2017 Emergency Classification and Coding
		Accident disaster	Coal mine accidents, fire accidents, etc.
		Public health emergency	Infectious diseases, food, drug safety events, etc.
		Social security emergency	Mass incidents, major criminal cases, etc.
	Disaster factors	Matter	Dangerous chemicals, etc.
		Energy	Water, electricity, fire, etc.
		Information	Rumors, etc.
	Time	Start time	Hour/minute/second–Day/month/year
		End time	Hour/minute /second–Day/month/year
		Duration	Minutes, hours
	Location	Administrative area	Town, country, city, province
		Longitude and latitude	Longitude, latitude
	Loss	Economic loss	Direct economic loss, indirect economic loss
		Personnel loss	Injury, death, relocation
		Other losses	The affected area of crops, the affected area covered
	Level	Level	Grade I (especially important), Grade II (important), Grade III (large), grade IV (general)
		Color	Blue, yellow, orange, red

,

Table 3. Concept level of hazard-affected carriers.

Primary Concepts	Secondary Concepts	Tertiary Concepts	Reference
Hazard-affected carriers	Social security emergency	Mass incidents, major criminal cases, etc.	GB T 32572-2016 Classification and Coding for Natural Disaster Exposure
	Disaster-stricken groups	Victims, affected people, etc.
	Property	Fixed property, current assets, etc.
	Resources and environment	Land resources, mineral resources, etc.

,

Table 4. Concept level of the emergency procedures.

Primary Concepts	Secondary Concepts
Emergency procedures	Preparedness
	Monitoring and warning
	Response
	Recovery

,

Table 5. Concept level of the measures.

Primary Concepts	Secondary Concepts	Tertiary Concepts	Emergency Procedures	Reference
Measures	Risk Management	Risk identification, risk analysis, etc.	Preparedness	Emergency Response Law of the People’s Republic of China
	Drill training	Emergency knowledge training, safety knowledge training, etc.
	Material reserve	Emergency material supervision, production, storage, allocation, etc.
	Monitor	On duty, real-time monitoring, etc.	Monitoring and warning	The National Response Plan
	Warning	Warning release, cancellation, etc.
	Command and Judge	Command setting, command operation, etc.	Response	NIMS
	Action Rescue	Information reporting, evacuation, vigilance isolation, etc.
	Emergency support	Material security, communication, information security, etc.
	Financial administration	Leasing matters, analysis of emergency costs, etc.
	Survey evaluation	Statistical survey, disaster assessment, etc.	Recovery	Emergency Response Law of the People’s Republic of China
	Recovery	Restoration of social order, restoration of public facilities, etc.
	Rescue	Compensation, pension, resettlement, etc.

,

Table 6. Concept level of the emergency supplies.

Primary Concepts	Secondary Concepts	Tertiary Concepts	Reference
Emergency supplies	Basic living supplies	Vegetables, fruits, etc.	GBT 38565-2020 Classification and code of emergency supplies
	Emergency equipment and supporting supplies	Explosion-proof gloves, diving gloves, etc.
	Engineering materials and machining equipment	Pipeline testing equipment, site monitoring equipment, etc.

and

Table 7. Concept level of the reference.

Primary Concepts	Secondary Concepts
Reference	Laws and regulation
	Policy document
	Emergency plan
	National standard
	Industry Standard
	Local standard
	Group standard

.

The subject mainly refers to the organizations and units involved in emergency management. According to the application requirements, the concept level of the subject is divided and shown in Table 1.

The object includes emergency type, disaster factors, time, location, loss, and level. The object concept is divided into four levels. The classification of the emergency type is based on the “Emergency Classification and Coding” [32], as shown in Table 2.

The classification of hazard-affected carriers is based on the “Classification and Coding for Natural Disaster Exposure” [33], as shown in Table 3.

The four emergency procedures include preparedness, monitoring and warning, response, and recovery. The conceptual hierarchy is shown in Figure 4.

Emergency measures are the measures that need to be taken in response to emergencies. They are divided according to the “Emergency Response Law of the People’s Republic of China” [34] and plan. To relate them with procedures, a procedure column is added, as shown in Table 5.

“Emergency supplies” refers to the materials and facilities. They need to be used in the response procedures and play a guaranteed role. According to the “Classification and code of emergency supplies” [35], the concept level the emergency supplies is divided, as shown in Table 6.

Decision making needs to be based on reference to avoid ambiguity of indicator content and other issues. For standard fine-grained knowledge, it is necessary to support them. Therefore, standard references play an essential role in standard digitalization. The standard reference mainly refers to normative documents, and the classification is shown in Table 7.

Relation Design

The KG contains two critical elements: named entities and relations. The relations form the semantic network of the KG. Relations include hierarchical relations and non-hierarchical relations. Hierarchical relations are the broader and narrower relationships between logical records. Hierarchical relations generally contain the relations between concepts and other concepts, as well as concepts and entities. Non-hierarchical relations are multiple simultaneous relationships obtained through practice without imposing a hierarchical structure. The common relations between concepts include part-of, kind-of, and instance-of. Combined with the conceptual hierarchy, the conceptual relation in the emergency response domain is designed and shown in

Table 8. The conceptual relation.

Concept	Relation	Concept
Subject	Issue/Put forward/Draft/Centralized in	reference
Subject	Use/Provide	Emergency supplies
Subject	Be responsible for/Support/Adopt/Provide/Participate in/Implement/Execute	Measures
Emergency supplies/object/emergency procedures/Measures	According to	reference
Object	Stage is	emergency procedures
Emergency supplies	Guarantee/Reserve	Measures
Object	Make… Affected	Hazard-affected carriers

.

The Construction of Emergency Response Domain Ontology

This paper uses the seven-step method to build emergency response domain ontology. The seven-step method is a classic ontology construction method proposed by researchers at Stanford University. It can use RDF/OWL to represent the ontology well. We design conceptual hierarchy and extract the relation between concepts. After that, generic relations are constructed between the concepts, and the instance of the concept is finally filled. This paper defines the emergency response domain ontology model as a five-tuple, which is shown in Equation (1).

$$\left(CO, RE, AT, RU,INS\right)$$

(1)

CO denotes the concept; RE denotes the relation set; AT denotes the attribute; RU denotes the inference rule; INS denotes the instance.

Combined with the concepts and relations, the emergency response domain ontology model is designed and is shown in Figure 4.

In Figure 4, the seven color modules represent seven primary concepts. The purple module represents reference, the yellow module represents the subject, the orange module represents emergency supplies, the green module represents emergency measures, the blue module represents emergency procedures, the pink module represents the object, and the gray module represents hazard-affected carriers. Each color module contains underlying concepts, attributes, and relations. The rectangular box represents the concept. The ellipse represents the attribute and the diamond represents the relation. The arrow points to the relation from the head concept to the tail concept.

3.2. Knowledge Extraction

As an essential part of the digitalization of ERS, knowledge extraction aims to meet actual emergency response needs. Through the digitization of ERS, we provide standard knowledge support for emergency decision making. It is necessary to extract knowledge from ERS and scenarios to obtain fine-grained knowledge. Then, the demand-based knowledge reorganization can be realized.

Standard knowledge extraction includes fine-grained knowledge extraction of standard content and coarse-grained knowledge extraction of standard structures. Standard fine-grained knowledge extraction takes standard content as the extraction object and the emergency response domain ontology model as the extraction basis. Equation (2) is the description model of standard fine-grained knowledge.

$$ERSK=ERS\left(sub,obj,car,pro,mea,sup,ref\right)$$

(2)

Sub, obj, car, pro, mea, sup, and ref denote the subject, object, hazard-affected carriers, emergency procedures, measures, emergency supplies, and reference, respectively. The seven-dimensional model includes subject, object, emergency procedures, measures, etc. The entities can be extracted based on these dimensions. Relations can be extracted based on Table 8. The emergency response domain ontology model covers the required ERS knowledge from seven conceptual levels. Thus, standard fine-grained knowledge extraction can be achieved.

Coarse-grained knowledge extraction takes a standard structure as the extraction object. ERS documents have fixed components, which are common elements in the standard, including cover, preface, introduction, scope, normative references, terms and definitions, chapters, etc. For extraction of common elements, XML tag parsing or information extraction methods can be used. Yang et al. [36] use XML tags to label the standard and extract common elements, including the introduction, scope, and chapter title.

In the life cycle of an emergency, the emergency response information requirements change with the emergency procedures. The purpose of the digitalization of ERS is to find corresponding response information and provide standard information support for emergency responses according to the different stages of emergencies. Therefore, we need to extract additional scenario information at different stages of the emergency to determine emergency tasks so that we can find decision indicators in ERS. The extraction of emergency information is the basis of knowledge reorganization and decision support. The description of an emergency includes time, location, type, loss, etc. It is necessary to determine the response requirements in a timely way as emergencies occur. For example, according to the type and damage of an emergency, the level can be determined according to ERS knowledge. Then, the response measures can be initiated, and emergency supplies can be provided.

The general emergency ontology models, such as ABC Ontology and MEO Ontology, contain the concepts of type, level, time, location, etc. The emergency response domain ontology model in this paper covers the whole emergency procedures, subject, and object. The object includes level, time, location, and type, consistent with the emergency description and the general emergency ontology model. Therefore, according to the seven-dimensional model, a description model of emergency scenario knowledge can be built and is shown in Equation (3).

$$ESK=ER\left(sub,obj,pro\right)$$

(3)

sub, obj, and pro denote subject, object, and emergency procedures, respectively. Knowledge extraction is carried out for emergency scenarios based on the emergency response domain ontology model. By extracting scenario descriptions, fine-grained knowledge at different stages of the scenario can be obtained. Then we can find the corresponding indicators in ERS and realize standard scenario knowledge reorganization.

For ERS and emergency scenario fine-grained knowledge extraction, methods related to information extraction in deep learning can be used. Generally, knowledge extraction can be divided into three steps as shown in Figure 5.

• Step 1: Data preprocessing. The format of standard documents and scenario cases is not standardized. It is necessary to deal with the unstructured format of a text dataset when developing models that use off-the-shelf NLP tools [37]. So, we should correct typos, delete duplicate paragraphs, delete blank lines, and so on.
• Step 2: Data annotation. To realize knowledge extraction accurately, it is necessary to label standard content and scenario content according to emergency response domain ontology. Common annotation forms include BIO, BIOE, etc.
• Step 3: Knowledge extraction. Based on data annotation, the entities, relations, and attributes can be extracted as triples. The common extraction models include CNN-based models, RNN-based models, etc. We can use a pipeline or joint method to realize the entity and relation extraction.

3.3. Knowledge Reorganization

Emergencies have characteristics of complexity, spread, and secondary disasters. In the response procedures, it is necessary to deal with the whole emergency process before, during, and after the different stages of the emergency. In this process, the ERS plays a vital role in standardizing the general technology, management, and technical requirements. Standard digitalization needs to realize the transformation from standard documents to fine-grained knowledge. It is even more necessary to learn the matching of standard knowledge to emergency response procedures and complete the reorganization of standard knowledge for scenario response. Knowledge reorganization combines knowledge and the construction of semantics, including ERS knowledge reorganization and standard knowledge reorganization for scenario response.

3.3.1. ERS Knowledge Reorganization

The ERS knowledge reorganization combines fine-grained knowledge according to the standard original structure. The ERS knowledge reorganization model (ERSKRM) is defined as Equation (4).

$$ERSKRM=\left(ERS\left(sub,obj,car,pro,mea,sup,ref\right),RE\left(ers\right),AT(ers\right))$$

(4)

ERS denotes emergency response standard; sub, obj, car, pro, mea, sup, and ref denote subject, object, hazard-affected carriers, emergency procedures, measures, emergency supplies, and reference, respectively; RE represents the relation set of ERS; AT denotes attribute. The results of knowledge extraction in ERS are expressed as triples. The combination of entities and relations constitutes the standard knowledge module. The reassembly of modules forms a standard part. The process is shown in Figure 6.

3.3.2. Scenario Response Standard Knowledge Reorganization

The standard knowledge reorganization for scenario response involves reorganizing the dismantled standard content according to scenario response requirements. The standard can be implemented in the process and provide an indicator for emergencies.

Standard knowledge reorganization for scenario response is based on emergency scenario knowledge and standard knowledge. The model of scenario response standard knowledge reorganization (SRSKRM) is defined as Equation (5).

$$SRSKRM=\left(ERSK,RE\left(ers,es\right),ESK\right)$$

(5)

ERSK denotes standard knowledge; ESK denotes emergency scenario knowledge; RE (ers, es) denotes the connection between ERS and ES. The SRSKRM is shown in Figure 7.

The process of standard knowledge reorganization for scenario response is divided into the following steps.

• Firstly, ERS knowledge and emergency scenario knowledge are extracted according to the emergency response domain ontology model.
• Secondly, retrieve ERS knowledge according to scenario knowledge requirements. We let scenario knowledge requirements interact with ERS knowledge to obtain valuable knowledge and discard useless ERS knowledge.
• Thirdly, the triples of ERS knowledge, demand, and result are effectively fused and judged to form a standard clause.
• Finally, different clauses are sorted out to form a standard knowledge module based on the scenario, which can provide indicator recommendations for emergency scenarios.

4. An Empirical Analysis

Based on the ERSDMM, this paper takes “Emergency Shelter for Earthquake Disasters Site and Its Facilities” (GB 21734-2008) as an example to show the process of digitalization of ERS. GB 21734-2008 can be divided into ten parts, including cover, preface, introduction, scope, normative references, terms and definitions, classification, site requirements, facility configuration, and other requirements.

4.1. Knowledge Extraction

4.1.1. Coarse-Grained Knowledge Extraction of ERS

The coarse-grained knowledge extraction of a standard structure is based on the standard structure to extract the common content and chapter titles. For example, Chapter 4 of GB 21734-2008 is Classification. See the following table for specific chapters. The content is shown in

Table 9. Content of Chapter 4 Classification.

Chapter Title

Content

4 Classification

Emergency shelter for earthquake disasters contains three classes —Class I emergency shelter for earthquake disaster, has comprehensive facility configuration and can accommodate the assisted personnel for more than 30 days —Class II emergency shelter for earthquake disaster, has general facility configuration and can accommodate the assisted personnel for 10 days~30 days —Class III emergency shelter for earthquake disaster, has base facility configuration and can accommodate the assisted personnel within 10 days

.

We label the standard structure with the XML tags reference “Tagging framework-based XML structurizing of national standard” in China, as shown in Figure 8. According to the XML tag, the chapter titles parse to show the composition of the standard structure.

<tech_item> denotes the first chapter item; <tech_itid> denotes the chapter id; <tech_itname> denotes chapter name; <tech_paragraph> denotes the second chapter item; <tech_part> denotes the third chapter item; <tech_ptbody> denotes the content item.

4.1.2. Fine-Grained Knowledge Extraction

It can be seen in Section 4.1.1 that standard structure knowledge extraction only obtains coarse-grained knowledge, and cannot obtain content fine-grained knowledge. Standard digitalization is not achieved. Therefore, it is necessary to extract standard fine-grained knowledge. Standard content knowledge extraction is based on the emergency response domain ontology model. The ontology model contains seven primary concepts and 38 secondary concepts.

Taking the Chapter 4 classification as an example, the extraction results are shown in

Table 10. Extraction results of Chapter 4 classification.

Content

Results

Emergency shelter for earthquake disasters contains three classes —Class I emergency shelter for earthquake disaster, has comprehensive facility configuration and can accommodate the recipients for more than 30 days —Class II emergency shelter for earthquake disaster, has general facility configuration and can accommodate the recipients for 10 d~30 d —Class III emergency shelter for earthquake disaster, has base facility configuration and can accommodate the recipients within 10 days

<Emergency shelter for earthquake disasters, contains, Class I emergency shelter for earthquake disaster> <Emergency shelter for earthquake disasters, contains, Class II emergency shelter for earthquake disaster> <Emergency shelter for earthquake disasters, contains, Class III emergency shelter for earthquake disaster> <Class I emergency shelter for earthquake disaster, has, comprehensive facility configuration> <Class II emergency shelter for earthquake disaster, has, general facility configuration> <Class III emergency shelter for earthquake disaster, has, base facility configuration> <Class I emergency shelter for earthquake disaster, accommodate, recipients> <Class II emergency shelter for earthquake disaster, accommodate, recipients> <Class III emergency shelter for earthquake disaster, accommodate, recipients> <Class I emergency shelter for earthquake disaster, time_is, more than 30 days> <Class II emergency shelter for earthquake disaster, time_is, 10 days~30 days> <Class III emergency shelter for earthquake disaster, time_is, within 10 days>

.

4.2. Knowledge Reorganization

4.2.1. ERS Knowledge Reorganization

Chapter 4 (Classification) in the standard shows that emergency shelters for earthquake disasters are divided into three categories. Class III emergency shelter for earthquake disasters has a base facility configuration. When we want to know what is included in the comprehensive facility configuration, we can only find it in Chapter 6 (Setting Configuration) by browsing the full text. After the standard digitalization, the knowledge related to the base facility configuration is reorganized through the relation to form the base facility configuration module, as shown in Figure 9. We can view all the knowledge related to the base facility configuration. This solves the problems of users’ difficulties with search and reading.

4.2.2. Scenario Response Standard Knowledge Reorganization

The purpose of scenario response standard knowledge reorganization is to solve the problem of rapid emergency response and provide indicator support. For example, the emergency scenario is described as follows: a magnitude 5 earthquake occurs in Sichuan province and the affected people need to be resettled for 7 days. Then we can obtain scenario knowledge: <Sichuan province, occurs, an earthquake>, <earthquake, level_is, magnitude 4>, <Sichuan province, resettle, recipients>, <Resettlement time, is, 7days>. Through these knowledge, corresponding information can be found in the standard knowledge graph and combined into indicators. The process is shown in Figure 10.

5. Discussion

5.1. Innovation

The standards in the emergency response field have the characteristics of solid professionalism, high complexity, and involvement of many subjects. In practical applications, they cannot meet the requirements of rapidity and accuracy of emergency responses. Standard digitalization can realize the mapping and fusion between standard knowledge, which is vital in responding to user needs. The main innovations of this paper are as follows.

• This paper proposes ERSDMM, which provides an implementation framework for the digitalization of ERS in the emergency domain.
• This paper proposes a seven-dimensional model based on the triangle model, and constructs an emergency response domain ontology model containing ERS knowledge and emergency scenario knowledge. The knowledge coverage of the emergency response domain ontology model is more comprehensive.
• Based on the ERS fine-grained knowledge and scenario knowledge, a reorganization model of the emergency scenario response standard is proposed, which provides an index basis for forming response decisions that meet the requirements of the standards.

This paper proposes an emergency response standard digitalization model. However, the research on knowledge management in the standard field still needs further improvement. By analyzing the standard documents, it is found that there are many types of ERS, and the standards contain rich charts. It is necessary to further supplement and improve the standard knowledge in KG in the future.

5.2. Research Prospects

Knowledge graph is an efficient knowledge management tool recognized in academia and industry [38]. Standards can effectively coordinate and standardize the operation of the industry in the digital age [39]. This paper proposes a standard digitalization modeling method for emergency response based on knowledge graphs. Based on this paper, future research on standard digitalization can focus on the following aspects.

Standard establishment: This paper proposes a knowledge reorganization model. This model can provide an indicator basis for decision making, but it does not realize the establishment of follow-up standards and other documents. The knowledge reorganization model needs to be studied further. By introducing the method of knowledge reorganization, the rapid formation of a standard or plan for emergency scenarios is realized.

Information system construction: the ERSDMM can be used to establish a KG and build a standard digitalization system. It contains ERS knowledge and scenario knowledge. The information fusion and integration of ERS knowledge, emergency state, disaster environment, and rescue resources are realized using information fusion and rule matching. The system can not only update the information but also recognize the flow and intelligence of crisis response and event disposal.

6. Conclusions

This paper proposes a standard digitalization modeling method for emergency response based on knowledge graphs. It provides a realization path for standard digitalization and solves the problems in the process of standard storage and application, which have vital practical significance.

This paper makes improvements in the method and application of emergency response standard digitalization and proposes a new standard digitalization method. Traditional standard digitalization is based on standard bibliography and standard full texts. The purpose is to achieve a search for standard full text. In this paper, standard digitalization takes the fine-grained knowledge of standard content as the data basis and realizes the expression and storage of standard fine-grained knowledge. It can quickly find the knowledge in the standard through demand matching and further realize the operation of the machine. This paper acknowledges the comprehensive and accurate utilization of standard knowledge. It designs a knowledge reorganization model. It can recognize the correlation and matching of emergency scenario knowledge and ERS knowledge and provide a decision-making basis for the occurrence of factual scenarios.