A Civil Aviation Customer Service Ontology and Its Applications

With the development of the civil aircraft manufacturing industry, foreign leading enterprises have gradually formed a trinity of “manufacture-market-service” organizational strategy. When choosing an aircraft manufacturer, customers always consider many key factors, in which the customer service capability plays a crucial role. From the perspective of knowledge, customer service is a process of knowledge re-creation. From the perspective of value chain, the value of customer service is to not only improve service efficiency and level, but also create a win-win market for manufacturers and airlines. Under this background, China's civil aviation customer service work also gradually receives attention. The enterprises led by Commercial Aircraft Corporation of China, Ltd^① (COMAC) gradually establish and optimize the civil aviation customer service system.

In the past years, we participated in the project “Developing the C919 Large Aircraft Customer Service Intelligence System” cooperated with COMAC. This project aims to solve information collection, information organization, intelligence service, and intelligence process construction in COMAC, in order to provide supports to the civil aviation customer service system building. However, during this project, we find several problems in real-world applications, including 1) inefficiency in collecting massive Web information, 2) high heterogeneity of complex data, 3) incomplete information analysis capabilities, and 4) inability to meet the requirements of enterprise intelligence decision making. Thus, how to standardize the knowledge in the domain of civil aviation customer service, to further provide effective intelligence services for enterprises is worthy to study.

In this paper, we introduce the idea of ontology modeling into civil aviation customer service, which tries to build a civil aviation customer service ontology. It could benefit to standardize heterogeneous data, discover implicit knowledge, improve the quality and efficiency of intelligence analysis, realize information sharing, and ultimately provides better intelligence services for the development of the civil aviation customer service system and intelligence personnel in enterprises. The built civil aviation customer service ontology^② has been published in OpenKG, which is the largest Chinese open knowledge graph platform.

Ontology building is to represent the main topological relationships that exists between entities in the building domain. This research topic has been studied for many years, and various ontologies [1, 2, 3, 4] have been constructed and published on the Web to facilitate the development of Semantic Web.

In the field of civil aviation, there are some research achievements in ontology building according to literature review. However, none of them focus on studying building civil aviation customer service ontology. Ma et al. [5] studied ontology building, fusion, mapping, and evolution in the aviation product domain. It primarily focuses on the aircraft itself, especially military products, and aims to achieve better information management and knowledge discovery by integrating, mapping, and evolving different ontologies. However, this approach of ontology engineering is not suitable for intelligence analysis in the domain of civil aviation due to significant differences between civil aviation and military aviation products.

Wang et al. [6] conducted research on the application of description logic [7] in the civil aviation ontology. The main focus is on using Description Logic to describe the ontology, performing reasoning, uncovering implicit hierarchical structures of concepts, identifying inconsistencies, and making necessary modifications to enhance the reasoning capability of civil aviation knowledge. This research has useful reference significance for reasoning over the civil aviation customer service ontology. Additionally, Wang et al. [8] also studied ontological relation extraction in the domain of civil aviation emergency management, achieving notable results in the application of emergency handling. However, our paper primarily focuses on building an ontology in the domain of customer service, analyzing the macro environment, key technologies, and competitors, in order to extract hidden intelligence information.

Zhang et al. [9] designed an ontology-based semantic information system for aviation products. Its main focus is on extracting structured model information of aviation products and leveraging semantic annotation to enrich the semantic information system. However, the research on ontology-related reasoning for aviation products was not conducted in their study.

Cheng et al. [10] built an ontology model for the civil aviation safety management and make initial efforts in building a knowledge graph for civil aviation safety management. The research primarily provides valuable insights on the extraction of structured, semi-structured, and unstructured data in this domain of civil aviation safety management. However, the ontology developed in the study does not adequately meet the intelligence decision-making requirements of the enterprises in the domain of civil aviation customer service.

The Enterprise Ontology [11] is a research project collaborated by the researchers from Artificial Intelligence Application Institute at the University of Edinburgh, IBM, and Unilever. This ontology models processes, planning, organization, strategy, marketing, and time, which are good modeling references for us to build the civil aviation customer service ontology. Thus, we choose the Enterprise Ontology as the reference ontology to build the the enterprise section of our civil aviation customer service ontology.

The purpose of ontology building in this paper has two aspects. Firstly, it aims to utilize the reasoning mechanisms of the civil aviation customer service ontology to uncover implicit intelligence information. By leveraging the reasoning capabilities of the ontology, a significant amount of knowledge that cannot be directly collected can be obtained during the phase of information processing, providing technical support for intelligence analysis for enterprise users. Secondly, it aims to address the issue of data heterogeneity to realize information sharing. Standardized and formalized ontologies can effectively describe complex relationships between entities, thereby providing models and methods for information organization and sharing.

We follow the five widely recognized basic principles proposed by Gruber [12] and refer to the W3C recommendations, to build our ontology. The principles for ontology building are based on the characteristics of the civil aircraft customer service ontology requirements, mainly including:

• Clarity. The definitions of classes and properties in the civil aviation customer service should express the characteristics of the domain of civil aviation customer service and cover relevant elements as many as possible. For example, the ontology defines the class Service, which includes sub-classes such as Engineering_Support_Service, Maintenance_Support_Service, Spares_Support_ Service, Customer_Training_Service, Flight_Operation_Support_Service, Technical_ Publications_Service, Market&Customer_Support_Service, and Digital_Customer_Service. Additionally, it establishes relevant properties such as Provide_Service, Has_Substitute, Use_ Technology, covering all the types of services and relationships required by aviation companies. It is of significant importance to objectively and accurately describe and represent the relevant elements of the ontology for subsequent reasoning and the discovery of implicit intelligence.

• Coherence. The definitions of classes in the civil aviation customer service ontology should form a mutually consistent and non-conflict semantic structural model. For example, when defining the class Aircraft, it explicitly states that the passenger capacity of three types of aircraft: Wide_body_ Aircraft (More than 200 seats), Medium_sized_Aircraft (100-200 seats), and Small_Airplane (Less than 100 seats). Any aircraft model can be an instance of only one type, avoiding conflicts and overlaps.

• Extendibility. Ontology building is an iterative design process of modification and gradual perfection. When adding general or domain-specific terms to the civil aviation customer service ontology, there is no need to modify existing class definitions. For example, for the Rule class, if a new regulation specific to the civil aviation domain is added, there is no need to modify the relevant definition of the class Rule.

• Minimal encoding bias. The ontology model describes the knowledge at the semantic level and is independent of the syntactic encoding methods. When constructing an ontology for the civil aviation domain, the actual system may employ different knowledge representation methods. Therefore, the representation and description of classes should not depend on a specific symbolic representation method, which benefit to reduce bias.

• Minimal ontological commitment. The ontology built in this paper is designed to meet specific knowledge sharing requirements without imposing excessive constraints on specific transactions. For example, the constraint on the object property Has_Technology of the class Corporation is set to Only, ensuring that the range of the Has_Technology relationship can only be the class Technology.

Besides the above principles, discovering implicit intelligence is another important target to build the civil aviation customer service ontology. Thus, in order to design rational reasoning mechanism, we need to accurately define the various properties of the classes in the civil aviation customer service domain and the relationship information of the relevant classes. In addition, too much redundant information will make the built ontology complicated, especially when the various classes of the other resources in the civil aviation customer service domain are overly described, which will lead to the ontology framework being biased from the civil aviation customer service subject to other aspects. Therefore, in the process of building the civil aviation customer service ontology, it is necessary to select the information of various properties in the civil aviation customer service domain with the concept of the domain as the center, in order to be concise, avoid unnecessary redundancy, and facilitate the operability of mining.

In this paper, we select the widely used seven-step method [13] to build the civil aviation customer service ontology. The seven-step method is a development method based on the ontology development tool Protégé [14], which has clear building steps, is practical, and has the highest maturity. Ontologies like the AsdKB Ontology [1] and CKGG ontology [2] are both developed following the seven-step method. The detailed building process is as follows:

1. Determine the domain and scope of the ontology. The primary objective of building the ontology in this paper is to discover implicit intelligence information and provide support for enterprise intelligence services. This paper focuses on three important aspects related to competitive intelligence systems, including environmental monitoring, technology tracking, and competitor analysis. The civil aviation customer service ontology aims to collect relevant information about the macro environment, industry environment, competitors, and customers from various sources such as enterprise websites, online resources, and commercial databases. The ontology provides valuable insights to support enterprise market decisions, and it will be used by enterprises to enhance their intelligence services and provide support for market analysis and decision-making. As for maintenance, it will be the responsibility of the organization or team involved in the management of the ontology to ensure its accuracy, relevancy, and currency.

2. Consider reusing existing ontologies. We reuse the standard RDF, RDFS, and OWL vocabularies, including rdf:type for linking instances to classes, rdfs:label for expressing labels of classes and properties, rdfs:subClassOf for describing class hierarchies, and rdfs:domain and rdfs:range for specifying instances of one or more classes for resources and the values of properties, respectively.

3. Enumerate important terms in the ontology. Before ontology building, we have built the thesaurus of civil aviation customer service, which contains a large scale of professional terms, and such terms do provide a solid foundation for ontology building. Important terms in the domain of civil aviation customer service such as Manufacturer, Supplier, Airline and Rule may be classified as classes. The terms like Employee, Marketshare, Be_Competitor, and Be_ Potential_Supplier can be classified as properties. During the process of enumerating terms in the civil aviation customer service ontology, we conduct research on the work requirements of various departments of COMAC, and according to the expert opinions and referenced relevant materials such as the Universal Decimal Classification, Chinese Library Classification, and standard literature subject thesauri, we finish this step. The results cover three main parts: customer service, competition analysis, and environmental intelligence. The customer service part contains customer training, spare parts support, maintenance services, and etc. The competition analysis primarily focuses on five companies, including Boeing, Airbus, Bombardier, and etc. The environmental intelligence part covers the areas such as technology, politics, society, and economy. Some examples are shown in Table 1.

4. Define the classes and the class hierarchy. To create classes and their hierarchical structure, we adopt a top-down development method, where we first define the most general classes in the target domain and then specialize them further. We have created initial classes such as Aircraft and Service, and then further classify each class. For example, the class Medium_sized_Aircraft should belong to the class Aircraft.

5. Define the properties of classes—slots. Properties are used to describe the characteristics of classes and instances. After defining the classes and the hierarchy of classes, it is time to describe the internal structure of classes. After selecting classes from the list of terms, most of the remaining terms are properties. We match properties with classes, and define properties and their hierarchies.

6. Define the facets of the slots. We specify the value type by defining the rdfs:range for each property. The range of each property is either an XML Schema data type or a class. We also define synonymous classes of the classes in the civil aviation customer service ontology.

7. Create instances. Generating an instance of the class requires three steps. First, select the class to which the instance belongs; second, generate an instance of it; third, fill in the values of the properties. Since creating instances is not our focus in this paper, we actually do not finish this step for the civil aviation customer service ontology.

With the method mentioned above, an initial ontology for the civil aviation customer service has been constructed, as shown in Figure 1. It has 69 classes, 42 object properties, and 19 data properties in total. In the later stages, it can be supplemented and expanded according to specific requirements because ontology development is an iterative process. The ontology will continue to be refined and improved as users discover new problems and enrich their requirements.

The designed ontology is intended to serve the intelligence analysis in the domain of civil aviation customer service. The basic components of a Corporate Intelligence System (CIS) typically include environmental monitoring, technological tracking, competitor analysis, market warning, strategy formulation, and information security. Implementing all of these components would require a significant amount of time and manual works. Considering the company's intelligence needs, we have selected three important aspects, which are environmental monitoring (e.g., regulations, standards, and latest technologies), technological tracking (e.g., civil aviation customer service and key technologies), and competitor analysis (e.g., enterprise relationships and media activities). Finally, we build the core classes and properties in the following subsections.

In the civil aviation customer service ontology, the class Corporation is divided into the following classes from the perspective of the industry chain: Supplier, Manufacturer, Airline, and MRO_Provider. The hierarchy of such classes is depicted in Figure 2.

The class Corporation has data properties such as Name, Time, Feature, Telephone, Marketshare, Assets, Employee, and Description. It also has object properties such as Has_Trademark, Has_Intellectual_Property, Has_Patent, Has_Technology, and Has_Expert. There are object properties between corporations, such as Cooperative, Be_Competitor, Sell_Aircraft_To, Be_Potential_Supplier, and Be_Potential_Customer.

Every corporation will take relevant actions or measures when managing, promoting, recruiting, and strategizing, which will be unconsciously “exposed” in public information sources. In the process of enterprise intelligence work, timely pay attention to the series of activities taken by competitors can help discover new technologies, new strategies, and new trends of rivals, which provides an effective way for enterprises to adopt coping strategies and learn from experience.

With reference to the Activity class designed in the Enterprise ontology, and considering the business activities involved in the domain of civil aviation customer service, we classify the class Activity into two sub-classes: Organizational_Dynamics and Media_Dynamic.

The class Organizational_Dynamics primarily describes the behaviors related to the changes in the organizational structure of corporations. It includes sub-classes such as Be_On_The_Market, Merger, Consolidation, and Bankruptcy. The class Media_Dynamics primarily describes daily operational behaviors generated during the business process of corporations. It includes sub-classes such as Aircraft_ Orders, Aircraft_Deliveries, Aircraft_Flight_Test, Airshow, Advertisement, Exhibition, Exchange, Survey, News_Conference, and Recruitment. The above mentioned class hierarchy is shown in Figure 3.

The class Activity has data properties such as Name, Time, and Description. It also has object properties such as Locate_In, Activity_At, and Restraint_On. There are object properties between corporations and activities, such as Participate_In for participation in activities.

By examining the information on activities published by corporations, we can obtain insights into their latest developments and facilitate the inference of implicit intelligence.

As the scope of civil aviation customer service continues to expand, the new customer service idea has led to fundamental changes in the content of customer service, which now involves engineering support, maintenance support, aviation material support, training support, and other areas, throughout the entire process of aircraft production, sales, and use. The constructed hierarchy of on the class Service is depicted in Figure 4.

We summarize that the class Service can be divided into the following sub-classses, including Engineering_Support_Service, Maintenance_Support_Service, Spares_Support_Service, Customer_ Training_Service, Flight_Operation_Support_Service, Technical_Publications_Service, Market& Customer_Support_Service, and Digital_Customer_Service. The class Service has different data properties (e.g., Name) and object properties (e.g., Use_Technology and Has_Substitute).

Civil aircraft products can be classified based on different characteristics such as aircraft usage, engine type, number of engines, flight speed, and etc. According to the classification standards of the Civil Aviation Administration of China, aircrafts can be classified into three types, i.e., large, medium, and small aircrafts, based on their seating capacity. This classification strategy is simple and intuitive, and can meet the needs of different levels of customers in real life. The class Aircraft has three sub-classes, which are Wide_body_ Aircraft for large wide-body aircrafts, Medium_sized_Aircraft for medium-sized aircrafts, and Small_ Airplane for small aircrafts.

Large aircrafts refer to the planes with a seating capacity over 200, featuring the dual-aisle configuration. Medium-sized aircrafts refer to the single-aisle aircrafts with a seating capacity of 100-200. Small aircrafts refer to the planes with a seating capacity below 100, often used for regional flights. The hierarchy of the class Aircraft is illustrated in Figure 5.

The aircraft class has data properties such as Name, Time, Price, and Type, which define the basic information about the aircraft. It also has object properties like Be_Manufactured, Use_Technology, and Consume_Material.

In the domain of civil aviation customer service, the demand for products (i.e., aircrafts) comes from both direct customers (i.e., airlines or leasing companies) and indirect customers (i.e., passengers). There are certain commonalities between these two types of customers, and according to our analysis, the demand can be categorized into several aspects, which are seating capacity, safety, economy, comfort, reliability, and good services.

As shown in Figure 6, the class Requirement has six sub-classes: Seating_Capacity, Security, Economy, Reliability, Comfort, and Friendly_Service. Seating_Capacity is a criterion for the classification of civil aircrafts, and aircraft products with different capacities are designed for different levels of demand. Seating_ Capacity includes the sub-classes More_than_200, 100_to_200, and 100_or_less. Security is the primary consideration for customers to choose an airline to fly, and is also a fundamental requirement for aircraft operation, without which there is no value in the utilization of the product. Economy is the primary factor in designing civil aircrafts, and is one of the evaluation indicators for airlines to choose a manufacturer when purchasing an aircraft. Reliability is an inherent characteristic of civil aircrafts and is an important factor in determining flight safety and influencing customer choices. Comfort is the subjective feeling of passengers about the aircraft design, and is a subjective determinant that affects passengers’ choices on air transportation means. Friendly_Service is an important responsibility of the manufacturer to the products and customers. The quality of services plays an increasingly important role in the customer's choices on manufacturers. These factors are both customer requirements for aircraft products and important aspects for manufacturers to consider when designing their aircrafts.

In addition to the main classes mentioned above, the classes involved in the civil aviation customer service ontology still include Government, Person, Technology, Spares, Intellectual-Property, Location, Occupation, Rule, and Product_Service_List.

Object properties between classes include Introduce (indicating governments set rules), Has_Expert (denoting companies’ experts), Has_Technology (indicating companies’ technologies), Produce (denoting the production of spares by a supplier), Sell_Spares_To (indicating that a supplier sells spares to another entity), Bought_Spares_From (indicating the purchase of spares by a manufacturer from a supplier), and Use_Technology (indicating the utilization of technologies by an aircraft or service).

With the built civil aviation customer service ontology, we can apply reasoning techniques to dynamically track advanced key technology information, and mine potential competitive relationships and supply chain relationships, in order to realize the collection, mining and analysis of competitive intelligence. Here, we list different targets which can be realized by ontology reasoning using rules in the following subsections.

Technology competition strategy generally includes technology leading strategy, technology following strategy, and technology substitution strategy, which all require enterprises to have strong keen observation on technology, and strong technology innovation ability. By tracking the development trend of key technologies, on the one hand, we can discover the emergence of new technologies in the industry; on the other hand, we can prompt the enterprises to improve its own technological innovation capability while learning from advanced technologies, and improve the core competitiveness in manufacturing and services in the aircraft market. In the scenario of dynamically tracking advanced key technologies, ontology reasoning can provide dynamic information on competitors’ technologies, patent applications, and etc. The targets are listed as follows:

According to the real-world data about competitive environment and technologies, ontology reasoning can help mine potential competitive relationships and supply chain relationships, which benefits to identify potential competitors and suppliers for enterprises. The detailed targets are listed as follows:

Based on the six targets listed above, we create rules with Semantic Web Rule Language (SWRL) for reasoning. Figure 7 shows the SWRL rule examples corresponding to targets.

Suppose that we have real-world data, with pre-designed reasoning rules, we can infer implicit knowledge or facts. Here, we give three examples for different targets:

• Example 1: The Boeing Company holds a patent for an improved space vehicle payload adapter design, utilized for attaching multiple satellites to a space vehicle during launch. With the rule corresponding to T1 (see Figure 7), we can infer that the Boeing Company has the technology related to space vehicle launching and satellites.

• Example 2: Qingdao Sentury Tire Co., Ltd. (i.e., a supplier) produces aircraft tires, which belong to the list of spares (i.e., consumed materials) consumed by COMAC (i.e., a manufacturer), but COMAC has not purchased any product from this supplier (may by provided by other suppliers). Thus, based on the rule corresponding to T6, we can infer that Qingdao Sentury Tire Co.,Ltd. is a potential supplier of COMAC.

• Example 3: Indian Airlines ordered a medium-sized aircraft B737, i.e., the passenger capacity is of 100-200, and the manufacturer of this aircraft is Boeing. The passenger capacity of aircraft C919 manufactured by COMAC is also 100-200. Thus, with the rules corresponding to T4 and T7 respectively, we can infer that C919 is a substitute for B737, and Indian Airlines is a potential customer of COMAC.

To further apply ontology reasoning techniques, we design a competitive intelligence system (CIS) based on our civil aviation customer service ontology for COMAC. An expressive ontology can well reveal the complex relationships between information and provide a solution for the standardization of heterogeneous data, so that high-quality reasoning results can be acquired. In the CIS, intelligence analysis is the core component, and is the only way to realize the conversion from information to intelligence. This paper focuses on the application of ontology reasoning in the analysis and processing module of the CIS, aiming to solve the problem of mining potential demand information in the process of intelligence analysis.

In the above descriptions, the creation of SWRL rules and the reasoning mechanism have been analyzed, which are the keys to the application of ontology reasoning in intelligence analysis. According to the intelligence requirements of COMAC, combined with the development trend of COMAC in the aircraft manufacturing industry and market conditions, the ontology-supported civil aircraft customer service CIS applies ontology reasoning to competitive intelligence analysis, and the specific framework is shown in Figure 8.

The application framework mainly contains three parts: user input, reasoning process, and system output. The input is the user's intelligence requirements, the reasoning process is mainly based on the reasoning mechanism of SWRL, and the output is the reasoning results for intelligence analysis. The reasoning process is based on the intelligence analysis requirements and Jess rule engine. For example, in the traditional CIS, we can identify potential aviation material suppliers for the COMAC by retrieving their basic information, but multiple information retrieval and analysis are required to obtain the desired information. However, with the constructed ontology, reasoning can be used to quickly obtain implicit results.

Through this information derived by reasoning, intelligence staff can further analyze the reasoning results at a deeper level to support leaders to make decisions. For example, through the information of various organizations or media activities recently participated or held by competitors, it can be predicted that this competitor will have new initiatives in new product development, product promotion, product marketing, technology introduction, technology reformation, and etc. Then the coping strategy can be timely planned for the new trends of competitors.

In this paper, we analyze that current competitive intelligence analysis faces with the problems of inaccurate information collection in the era of big data, difficulty in unifying heterogeneous data, and insufficient functionality in intelligence analysis and processing. These limitations have led to the confusion and disappointment among enterprises regarding their purchased CIS products, which cannot meet the requirements of intelligence decision making. Thus, to solve the above problems, we build and publish the civil aviation customer service ontology, and apply ontology reasoning based on SWRL to support constructing a new CIS system in the domain of civil aviation customer service. By applying the ontology to the construction of CIS, this paper not only explores new possibilities for ontologies in the field of competitive intelligence but also provides valuable insights for the development of civil aviation customer service. In the future, we will consider to build the knowledge graph to further enhance the civil aviation customer service intelligence system. Besides, we plan to evaluate the performance of ontology supported CIS in the real-world civil aviation customer service applications.

M.L. and T.W. proposed the idea of this work and designed the method. X.C. finished initial experiments. M.L. and X.C. wrote the paper. T.W. and Y.L. revised this paper and gave a lot of suggestions. All authors read and approved the final manuscript.

This work is supported by the National Natural Science Foundation of China (Grant No. U21B6001, 62006040, 62376058, U21A20488), the Fundamental Research Funds for the Central Universities, and ZhiShan Young Scholar Program of Southeast University. We thank the Big Data Computing Center of Southeast University for providing the facility support on the numerical calculations in this paper.