Supply chain knowledge management supported by a simple knowledge organization system

Accepted Manuscript Supply chain knowledge management supported by a simple knowledge organization system Cristian A. Rodríguez-Enríquez, Giner Alor-Hernández, Jezreel MejiaMiranda, Jose L. Sánchez-Cervantes, Lisbeth Rodríguez-Mazahua, Cuauhtémoc Sánchez-Ramírez PII: DOI: Reference:

S1567-4223(16)30038-2 http://dx.doi.org/10.1016/j.elerap.2016.06.004 ELERAP 674

To appear in:

Electronic Commerce Research and Applications

Received Date: Revised Date: Accepted Date:

7 September 2015 30 June 2016 30 June 2016

Please cite this article as: C.A. Rodríguez-Enríquez, G. Alor-Hernández, J. Mejia-Miranda, J.L. Sánchez-Cervantes, L. Rodríguez-Mazahua, C. Sánchez-Ramírez, Supply chain knowledge management supported by a simple knowledge organization system, Electronic Commerce Research and Applications (2016), doi: http://dx.doi.org/ 10.1016/j.elerap.2016.06.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

SUPPLY CHAIN KNOWLEDGE MANAGEMENT SUPPORTED BY A SIMPLE KNOWLEDGE ORGANIZATION SYSTEM Cristian A. Rodríguez-Enríquez ([email protected]) Giner Alor-Hernández (corresponding author, [email protected]) Instituto Tecnológico de Orizaba Jezreel Mejia-Miranda ([email protected]) Centro de Investigación en Matemáticas Unidad Zacatecas Jose L Sánchez-Cervantes ([email protected]) Lisbeth Rodríguez-Mazahua ([email protected]) Cuauhtémoc Sánchez-Ramírez ([email protected]) Instituto Tecnológico de Orizaba Last revised: June 30, 2016 _____________________________________________________________________________________ ABSTRACT Supply chain management and business-to-business procurement present several drawbacks in terms of knowledge management. Every stage of a supply chain lacks an effective approach to integrate data workflows for knowledge acquisition. Thus, semantic technologies such as the simple knowledge organization system (SKOS) are being adapted to the requirements of the knowledge management systems of companies. The literature is focused on assets, data, and information elements of exchange among supply chain partners, even though improved integration and collaboration require more complex features of know-how and knowledge. This article proposes a new software architecture named SKOSCM to offer a brokerage service for e-procurement in supply chains. The approach uses ontologies and a Web-based platform that improves collaboration among supply chain partners. A case study is proposed in order to validate the software architecture’s development. Keywords: E-procurement; evaluation study; knowledge management; simple knowledge organization system; SKOS; semantic web; supply chain management _____________________________________________________________________________________ Acknowledgments. This work was sponsored by the National Council of Science and Technology (CONACYT), the National Technologic Institutes of Mexico (TecNM), and the Secretariat of Public Education (SEP) through PRODEP. _____________________________________________________________________________________

1

1. INTRODUCTION Since the efficient use of knowledge is critical to the survival and success of companies in competitive globalized markets, knowledge management systems (KMS) have become an essential tool for supply chain management (SCM). The importance of knowledge sharing relies on its strong potential for problem solving and enhancing organizational performance, decision-making, and innovation. In this context, knowledge management (KM) has emerged as a process for capturing, developing, sharing, and effectively using organizational knowledge (Davenport 1994). In addition to knowledge management, collaboration among supply chain (SC) partners is another promising research area for academics and practitioners. Companies can obtain several benefits for their supply chains from this collaboration. They include intelligent inventory management, new product development, and collaborative product design management, to mention but a few. There is a growing recognition that SCM gives companies significant opportunities to develop a strategic advantage toward their competitors (Wen and Gu 2014). This interest has steadily increased since the decade of 1980, when firms perceived the benefits of collaborative relationships within and beyond their own organizations. Firms recognized that they could no longer effectively compete by isolating themselves from their suppliers or other entities in the supply chain (Lummus and Vokurka 1999). Following the premises that knowledge is an asset for companies and that knowledge management and collaboration are essential in every echelon in the supply chain, several KMSs have been developed. They are available in the market and are suitable for e-procurement, supply chain management, knowledge management and collaboration among supply chain partners. However, they fail to detect issues at the procurement stage. Further, data transformation into knowledge depends on each supply chain administrator’s expertise, since the type of data generated is different in every supply chain echelon. The term supply chain is subject to different interpretations. Some are related to management processes, while others stress the structural organization of businesses. However, supply chain management integrates the management of supply and demand. Moreover, according to the Council of Supply Chain Man-

2 agement Professionals, it encompasses “the planning and management of all activities involved in sourcing and procurement, conversion and logistics.” Thus, supply chain management also covers coordination and collaboration with channel partners, such as customers, suppliers, distributors, and service providers. 1 This research focuses on the procurement stage, since a common source of problems in a supply chain is raw materials procurement. Raw materials and assets determine the success or failure of some subsequent production process. Also, the first concern of knowledge management is supplier selection. Note that knowledge management allows for the improvement of production processes, operational and organizational performance, and decision-making processes, among others. Considering the key elements for a successful supply chain, which are collaboration among all partners and the live sharing of data among them, the success of semantic Web technologies in B2B and e-commerce has turned them into promising technologies to design and build effective business collaboration systems and KMS in companies. According to Schandl and Blumauer (2010), a simple knowledge organization system (SKOS) has recently become one of the sweet spots in the Linked data ecosystem. It is the key to improving semantic information management thanks to the capabilities of taxonomy and thesaurus management, text mining and entity extraction, and knowledge engineering and ontology management. SKOS is an area of study developing specifications and standards to support knowledge organization systems (KOS) such as thesauri, classification schemes, subject heading systems, and taxonomies within the framework of semantic Web. However, even though several studies have reported theoretical and practical foundations for knowledge management (Lee and Goodwin 2006), little research has addressed the use of semantic Web technologies, such as SKOS for supply chain management. SKOS allows for the automatic information exchange across the supply chain among the production, procurement, and distribution stages. Also, it is produced in an automated way, due to domain-specific knowledge inferred from the ontology and passed over to SKOS to provide benefits for organizations.

According to Thomas and Griffin (1996), the three recognized, fundamental stages of the supply chain – procurement, production, and distribution – have been managed independently, buffered by large inventories. However, increasing competitive pressures and market globalization are now forcing firms to develop supply chains that can quickly respond to customers’ needs. As a result, to remain competitive, these firms must reduce operating costs while continuously improving customer service. 1

3 With recent advances in communication and information technologies (ICT), especially semantic Web technologies, organizations can reduce operating costs by coordinating these stages. Our aim is to introduce SKOS-based supply chain management (SKOSCM), a new semantic Webbased approach that uses SKOS to improve and facilitate knowledge management among all supply chain partners between the procurement and production stages. The importance of using a new approach for knowledge management relies on the benefits of the linked data paradigm and SKOS. In this regard, this research identifies opportunities such as open data sources of knowledge (linked open data, social networks, and others) and some benefits, including automation of data organization and procurement for nonexpert organizations. Our main contribution of this study is the application of knowledge management through SKOS for supplier selection, which matches with some preconditions to maintain a desired production in a multi-provider scenario for a milk supply chain as a proof of concept. Section 2 presents the state-of-the-art about Web technologies for knowledge management in supply chains and e-procurement. Section 3 discusses the general research methodology, presents the data acquisition method, and details how SKOS is used in supply chain management. It describes how SKOSCM resolves the problems and how it is employed in the case study. Section 4 discusses SKOSCM evaluation, addresses the features of KMS, and presents the evaluation finding. Finally, Section 5 summarizes results obtained from applying SKOSCM and provides concluding remarks and directions for future work. 2. THE STATE-OF-THE-ART FOR SUPPLY CHAIN SYSTEMS Several works have been recently proposed to develop e-procurement and supply chain management systems. Therefore, the relevant topics for this research include e-procurement, supply chain management, semantic technologies, and knowledge management. 2.1. Hierarchical Business Domain: Knowledge Acquisition by Using Ontologies Originally, the term ontology was rooted in philosophy where it denotes ‘‘the science of what is, the kinds and structures of objects, properties events, processes, and relations in every area of reality” (Floridi 2008). According to Gruber (1993), in the context of knowledge sharing an ontology is a specification of a

4 conceptualization. Ontologies can be a component of knowledge-based systems, but they also provide a common language for communication among domain analysts, developers, and users. In this research, we have improved SKOS with the use of ontologies. A brief comparative analysis of ontologies for supply chain management is performed over the results of the initial search query of “supply chain” and “ontology” throughout the online databases used for a keyword-based search. Additionally, alternative terms are employed to reflect the actual use of various terms for the key concepts. These terms are: “supply network,” “supply chain management,” “knowledge model,” “semantic model,” and “ontology model.” They are based on the parameters proposed by Scheuermann and Leukel (2014) in their literature review. The literature provides and discusses various SCM ontologies suitable for a range of industries and tasks. Table 1 lists these suitable ontologies. INSERT TABLE 1. ONTOLOGIES FOR SUPPLY CHAIN MANAGEMENT According to the literature, IDEON and SCOntology are suitable for the purpose of this research. They are used to design the SKOS-based software architecture for knowledge management. IDEON is an extensible ontology to design, integrate, and manage collaborative distributed enterprises. On the other hand, SCOntology is a formal approach toward a unified and integrated view of the supply chain. IDEON and SCOntology are appropriate for our research because they support production processes, activities, resources, product delivery, and return schemas. IDEON represents the business domain depicted in Fig. 1 and provides a hierarchical schema to organize and classify data among the supply chain data flows. This is appropriate for supply chains due to the highly-coupled relationships among supply chain partners. INSERT FIGURE 1. BUSINESS DOMAIN 2.2. Electronic Procurement The e-procurement supply chain includes indent management, e-tendering, e-auctioning, vendor management, catalogue management, purchase order integration, order status, ship notice, e-invoicing, epayment, and contract management. Therefore, communication among all supply chain partners is crucial to the success of these activities. One of the most suitable forms of improving communication is the use of an electronic procurement system, such as a brokerage service using semantic Web technologies (Alor-

5 Hernández et al. 2014), or t-based procurement solutions (Rita and Krapfel 2015). We will describe relevant works that developed different approaches to knowledge management. Several techniques about knowledge management at the e-procurement stage have been reported. Some of them include ontologies and semantic Web services (Koumoutsos and Thramboulidis 2009), software agents (Sun et al. 2012), case-based systems (Luu et al. 2003), and electronic agents (Hadikusumo et al. 2005). One of the most used is software agents (Lee et al. 2009). The success of knowledge management is not exclusive to e-procurement. As regards e-commerce, Wen (2007) introduced a knowledge-based intelligent e-commerce system (KIES) to sell agricultural products. The work demonstrated that knowledge-based systems visibly improved e-commerce systems. Also, in terms of coordination improvement, according to Rao et al. (2012), multi-attribute auction enables negotiation on several attributes, in addition to the price, that need coordination between supply and demand stages, such as quality, quantity, delivery time, and service levels. In the same context, Huang et al. (2013) discussed the design of a hybrid mechanism for eprocurement that implemented a multi-attribute combinatorial auction. Moreover, Huang et al. (2011) proposed a design mechanism for e-procurement auctions. These auctions require the participation of suppliers to provide data through a centralized system that tried to find the best bid (Pareto optimal). 2.3. Supply Chain and Knowledge Management Recent advances in knowledge management and semantic technologies have their emphasized for organizations using information technologies. Through electronic networks, these companies can achieve data integration by tightly coupling processes at the interfaces between every stage of the value chain. According to Williams (2008), electronic linkages in value chains have been fundamentally changing the nature of interorganizational relationships. There has been significant research in the SCM environment. By using ontologies, such as semantic Web technologies (Munoz et al. 2015), organizations are redesigning their internal structure and external relationships. Similarly, they are trying to create new knowledge networks to facilitate data communication, information, and knowledge management (Mylan et al 2014), while improving coordination (Wu 2001), decision making (Peiris et al. 2015, Börjeson et al.

6 2014), knowledge transfer (Becker and Zirpoli 2003, Holtbrügge and Berg 2004), knowledge sharing (Soliman et al. 2005, Ryoo and Kim 2015, Göksu et al. 2015), knowledge redundancy (Sivakumar and Roy 2004), knowledge exploitation (Soliman et al. 2005), knowledge discovery (Lau et al. 2009), and planning. With these organizational changes, flexible and scalable information technologies are preferred in order to take advantage of the know-how from supply chains. Since integration among supply chain partners is essential, Hjaila et al. (2016) concluded that tactical levels of decision-making in centralized supply chains helped improve the production stage by focusing on the efficient coordination between the supplier and the production echelons. Furthermore, Peng et al. (2016) determined how information technologies (IT) and knowledge acquisition (Hult et al. 2007) affected firms performance, whereas Soliman et al.( 2005) studied the relationships among supply chain, agility of firms, and knowledge management. Similarly, Wang et al. (2015) found that operational research tools could improve business decisions for the service supply chain. Several approaches include a knowledge-based framework for a dynamic re-configuration of supply chains over time (Piramuthu 2005). They suggest that the management of supply chain relationships is a means to share and acquire knowledge, instead of building external business relationships (Halley et al. 2010). For instance, Niemi et al. (2009) pointed out the process of knowledge accumulation – a process for data recollection with or without classification. They aimed to evaluate the adoption of complex practices in supply chain management (Niemi et al. 2010). In addition, other research analyzed the supply and value chains of different countries and businesses (García-Cáceres et al. 2014) to detect agents, phases, stages, and factors that influenced on every single echelon. Also, Avelar-Sosa et al. (2014) analyzed the effects of regional infrastructure over supply chain services performance in manufacturing companies. Finally, Espinal and Montoya (2009) identified the state-of-the-art and the current use of ICT in supply chains. They analyzed existing studies on the subject and eventually observed that most of these technologies contributed to cost reduction and improved the information flow among supply chain actors.

7 2.4. Additional Details Table 2 provides more details of the works above described. INSERT TABLE 2. LITERATURE REVIEW HIGHLIGHTS Even though the literature discusses several interesting approaches to knowledge management, many commercial tools can also handle this task. Since they are not reported in the analyzed works, we searched for them on the Web. Table 3 compares these tools by highlighting their features. INSERT TABLE 3. COMMERCIAL TOOLS FOR KNOWLEDGE MANAGEMENT Despite their significance, these initiatives have limitations. For instance, (1) most of these works are solely based on electronic data management, while knowledge management is usually addressed only for knowledge sharing (KS). Also, (2) those initiatives that allow for knowledge management do not provide user-friendly interpretations of knowledge inferred. Finally, (3) the system that is able to achieve knowledge management does not provide a platform for data acquisition at all levels in order to improve communication among supply chain partners. These deficiencies can be improved by: (1) designing an automatic knowledge management mechanism that transforms raw data into knowledge; (2) including a system that automatically brings the user a knowledge management tool by providing a set of friendly user interfaces; and (3) ensuring a straightforward form of interpreting knowledge acquired. This article addresses the aforementioned issues by proposing SKOSCM, a SKOS-based approach for knowledge management by using semantic Web technologies, due to their ability to communicate and share data with semantic meaning. 3. METHODOLOGY The research methodology comprises four stages: SKOS-based data acquisition, SKOSCM description, the case study, and an analytic hierarchy process (AHP) evaluation of the desirable KMS features. These stages were selected due to the following premises: (1) Data modeling provides an interpretation or insight about a real thing or phenomena (Hay 2013). A data model for knowledge management is required in order to obtain knowledge representation of data flows in a supply chain.

8 (2) The data model should be described in order to understand how knowledge is represented. (3) The design of a software solution is needed. In this regard, software architecture is an essential element of software system design and construction (Shaw and Clements 2006). It is important to recognize common paradigms, so that high-level relationships among systems can be understood and new systems can be built as variations of old systems (Taylor et al. 2009). (4) A case study should be carried out to understand the degree to which certain phenomena are present in a given group or how they vary across cases (Flyvbjerg 2006). 3.1. SKOS Model for Data Acquisition By using SKOS, concepts can be identified using a string of characters used to identify a name of a resource (URIs), labeled with lexical strings in one or more natural languages (for instance, business domain language), assigned with lexical code notation, documented with various types of notes, linked to other concepts, and organized into informal hierarchies and association networks, aggregated into concept schemes, grouped into labeled and/or ordered collections, and mapped to concepts in other schemes. All these features allow for knowledge integration and management in a particular domain, such as SCM. Fig. 2 depicts the major elements of the SKOS data model. INSERT FIGURE 2. SKOS DATA MODEL Features of the SKOS data model (Baker et al. 2013) are listed below by defining the elements depicted in Fig. 2. These features are: identifying, labeling, documenting, linking and mapping concepts (even in other schemas), and aggregating concepts into concept schemes or collections. Based on its features, SKOS has been used to express, in an interoperable way, different types of knowledge organization systems (specifically for a supply chain model). SKOS refers to a set of terms or concepts, whether listed with definitions (glossaries), in hierarchical structures (basic classifications or taxonomies), or characterized by more complex semantic relations (thesauri, subject heading lists, or other advanced structures). Data source. For this research, information was obtained from broker Web pages, ERPs, CRMs, and intranet applications, among others. It is important to mention that some of these applications are populated with data consulted from the U.S. Department of Agriculture (USDA) to meet the data volume required for the inference task of SKOS. Also note that the USDA data have been used in past JCR indexed articles. Fig. 3 illustrates a conceptual schema for data acquisition.

9 INSERT FIGURE 3. CONCEPTUAL SCHEMA FOR DATA ACQUISITION AND FEEDBACK Knowledge acquisition is elated to the following aspects to be improved in supply chain management: •

Demand. It is an essential element in supply chain management. It makes companies and their partners meet the needs of customers, rather than the production process.

•

Integration. Integrating supply chain processes helps every member reduce its inventory costs.

•

Collaboration. It strengthens relationships among members by improving teamwork and helping all members increase their business

•

Communication. Effective communication helps the entire supply chain improve the efficiency and productivity of its operations by allowing all members to share the same demand and operational information.

All data used by the SKOS engine are obtained from the Web-based platform SKOSCM. We next describe the SKOSCM architecture. 3.2. SKOSCM Architecture While knowledge management focuses on the supply side, knowledge creation focuses on demand. Johnson and Whang (2002) proposed the term e-collabor3ation for systems that facilitate Internet-based coordination of decisions across all supply chain members. In this research SKOSCM also concentrates on the demand through Internet-based coordination of knowledge. The use of the proposed technologies is intended to generate raw knowledge in an automated way. Moreover, SKOSCM relies on a layered design to organize its components. This layered design allows for scalability and easy maintenance, since its tasks and responsibilities are distributed. The general SKOSCM architecture is shown in Fig 4. Each of its components is explained in Appendix 1. INSERT FIGURE 4. SKOSCM ARCHITECTURE Carrer-Neto et al. (2012) and Ruiz-Martínez et al. (2012) used a top-level-ontology-based framework to populate biomedical ontologies from texts. Thus, a social knowledge-based recommender system can be an opportunity to improve SKOSCM to extract knowledge from text as a data source and the recommendation of knowledge resources through social network interactions among supply chain customers. This desired feature fits with the goal of automatizing data acquisition, and it requires the participation of the final echelon of the supply chain, the customer.

10 3.3. A Case Study of a Milk Supply Chain at the E-Procurement and Distribution Stages To illustrate the functionality of SKOSCM, we provide a case study for a multi-level scenario of eprocurement in a milk supply chain. This case study is intended to encourage supply chain practitioners and managers to develop and refine knowledge management by using semantic Web technologies. Thus, let us suppose that a supply chain in the domain of milk production in the U.S. requires information (knowledge) on time occurring between the procurement stage and distribution stage, since in some cases the product expires during its transport and represents losses and addition costs for the company. The eight stages in the milk supply chain are: •

Production of feed for cows. According to the Economic Research Service (Service 2007) of the U.S. Department of Agriculture (USDA), the dairy supply chain begins with growing crops such as corn, alfalfa, hay, and soybeans to feed dairy cows.

•

Milk production. According to the USDA National Agricultural Statistics Service, dairy cows are housed, fed, and milked in dairy farms across the country.

•

Milk transport. According to the USDA Agricultural Research Service, and as part of the University of Arkansas “Greenhouse gas emissions from milk production and consumption in the United States” research, milk is transported from the farm to the processing plants in insulated tanker trucks. The average truck carries 5,800 gallons of milk and travels approximately 500 miles in a round trip.

•

Processing. There are more than 1,000 U.S. processing plants that turn milk into cheese, yogurt, ice cream, powdered milk, and other products (USDA ERS).

•

Packaging. This stage is typically a responsibility of the dairy processor. Both paperboard and plastic containers are designed to keep dairy products fresh, clean, and wholesome.

•

Distribution. Distribution companies deliver dairy products from the processor to retailers, schools, and other outlets in refrigerated trucks.

•

Retail. According to the USDA Agricultural Marketing Service and the market trends published by the Progressive Grocer Magazine, milk and dairy products are available in 178,000 retail outlets of all shapes and sizes.

•

Consumer. As mentioned in the National Nutrient Database for Standard Reference Release 28, which is published and maintained by the USDA, milk and milk products deliver nine essential nutrients to consumers (calcium, vitamin A, vitamin D, vitamin B12, proteins, potassium, riboflavin, niacin, and phosphorus).

These stages must be grouped in the three main categories previously mentioned in the conceptual schema of Section 3. The groups are: (1) production (cows feed production and milk production); (2) pro-

11 curement (processing and packaging); and (3) distribution (distribution and retail). To illustrate the current supply chain cycle, Fig. 5 proposes a UML sequence diagram for the milk supply chain. INSERT Figure 5. UML Sequence Diagram of Milk Supply Chain Constraints for the case study to support the proof of concept. Certain constraints (C) and preconditions (P) must be considered for this system. SKOSCM is still at an early stage of development to demonstrate feasibility SKOS in knowledge management and insurance from the same raw data. •

P1. Cow feed production takes time. For this case study, SKOSCM requires only the final production of cattle feed and its nutrimental data.

•

P2. Time to deploy SKOSCM in a real milk supply chain: In this case study, deployment time was two weeks, during which data was collected to obtain results through SKOS.

•

C1. Raw data can be transformed into knowledge only with enough volume of them. This case study considers actual data from the USDA as historical data and a large volume of total information. Historical data are combined with raw data from a dairy farm.

•

C2. Benefits from using SKOSCM have only been tested for the production stage in the processing plant. A local dairy farmer from Mexico currently makes use of the tool at no cost, since SKOSCM is still at the experimentation stage with functional prototypes. Therefore, other stages of the supply chain must be tested.

Use of SKOSCM prototype. Once the SC echelons are identified, constraints for this proof of concept are described, and the milk production process is explained. We now describe the use of SKOSCM. First, Fig. 6 depicts the Web-based user interface to capture a cow feed (alfalfa) order request between the milk production echelon and the raw material supplier. Second, as can be observed in Fig. 7, once the product order request is captured, several suppliers inform their product availability, prices, and quality assets to the milk producer. The milk producer then selects the desired supplier (e.g., Baltic Cow Inc.) and confirms the purchase order of raw material (in this case, alfalfa); then, cows are fed. At the end of the day, as Fig. 8 depicts, the milk producer captures the amount of milk produced and the processing plants are notified about the available production. Then, these processing plants send the appropriate number of refrigerated of trucks to the milk producer in order to transport raw milk to their own silos. Finally, as can be seen in Fig. 9, the processing plant captures the final production of cheese, ice cream, and yogurt. INSERT FIGURE 6. THE CAPTURE OF A PURCHASE ORDER REQUEST

12 INSERT FIGURE 7. SELECTION OF A SUPPLIER INSERT FIGURE 8. THE CAPTURE OF MILK PRODUCTION INSERT FIGURE 9. CAPTURE OF THE FINAL PRODUCTION In all of these cases, the expected production of milk products reflects estimated values; so the processing cost and time may not be the same. Moreover, the packaging echelon expects more product than the actual stock, and consumer demand is not fulfilled. In this scenario, knowledge acquisition through the SKOS approach retrieves the following information: (1) Cow feed production increases or reduces milk production; (2) cow feed production affects the quality of milk produced; (3) due to a lack of communication between the processing and the packaging phases, milk remains longer in the packaging process. The packaging unit is not familiar with the dairy product that should be released first, since it depends on milk quality and the fact that certain derivate products cannot be produced under the same processing time. Poor or little communication affects operations at all stages in the milk supply chain. The degree of communication among all echelons in a supply chain is directly proportional to the time needed to take countermeasures to change the production process in early stages. To take advantage of knowledge generated in the supply chain, it is suitable to perform a more detailed analysis about production implications in the retail stage (sales and profits, among others). These data can be obtained with the SKOS engine depicted in Fig. 10. However, for a more accurate analysis, business analysts can use data mining and big data tools in conjunction with SKOS. The SKOSCM processes detailed in this case study can also be viewed in four demonstrational videos available at the SKOSCM website (SKOSCM 2015). INSERT FIGURE 10. RESULTS OF THE SKOSCM ENGINE After using SKOSCM, we noted that the milk supply chain sequence depicted in Figure 5 does not properly follow the three fundamental stages of every supply chain. “Cows feed production” was unbound from a procurement echelon (multi-level procurement), and the data workflow is broken. This hindered the live issue detection (before it happened) due to poor quality assets used at the production stage. After detecting that a procurement echelon is needed, the value chain can be modified to include three new stages corresponding to alfalfa supply. Fig. 11 depicts this sequence and shows how SKOS engine results ena-

13 ble detection of communication problems that affect data workflow among supply chain partners. INSERT FIGURE 11. UML SEQUENCE DIAGRAM OF EXTENDED MILK SUPPLY CHAIN Transformation of raw data into knowledge. The following material explains the labels (Miles and Bechhofer 2009) used in SKOSCM and their use for knowledge management. Also, the section describes data workflow to SKOS transformation, data into knowledge transformation, and how SKOS can be used with data exploitation for error detection. In KOSs, semantic relations play a crucial role in defining concepts. The meaning of a concept is defined by the natural language words in its labels, but also by its links to other concepts in the vocabulary (Isaac and Summers 2008). Mirroring the fundamental categories of relations that are used in vocabularies, such as thesauri, SKOS supplies three standard properties depicted in Table 4. INSERT TABLE 4. SKOS STANDARD PROPERTIES To assert that one concept is broader or more general in meaning than another the skos:broader property is used. Then, the skos:narrower property is employed to assert the inverse, when one concept is narrower and more specific in meaning than another. This feature describes relationships among supply chain partners. Finally, skos:related can be used to assert an associative relationship between two concepts. Note that ex:cows rdf:type skos:Concept; in line 1 of Table 5 defines a concept. In this case a “cow” is related to the “milk production” concept in line 4. In addition, as can be inferred from the previously described labels, every concept in SKOS is related to another. However, for a proper understanding of how SKOS operates in supply chains, Fig. 12 illustrates its work process in the milk supply chain. INSERT TABLE 5. SKOS RELATIONSHIP BETWEEN A COW AND MILK PRODUCTION INSERT FIGURE 12. SKOS EXAMPLE FOR MILK PRODUCTION Fig. 12 represents the relationship between cows feed and milk production. In natural language the diagram represents the following statements: (1) Cows are mammals, and mammals are animals; therefore, a cow is an animal. (2) Alfalfa is a plant, but it is also a feed crop. (3) For better quality milk, high-quality feed crop is needed for feeding the cows

14 The fact that a cow is an animal (first statement) is inferred from the relationship between mammals and animals by using the skos:broaderTransitive label. This accesses the transitive closure of a hierarchy expressed with skos:broader to determine the first and second statements. Also, in the third statement, a semantic Web inferencing tool obtains the relationship between milk production and alfalfa used to feed cows thanks to the skos:related label, which links ex:milkproduction with ex:feedcrops. SKOS is able to infer knowledge from relationships and raw data, but in the context of supply chain, it is important to understand how these relationships can be translated. Fig. 13 thus introduces a causal loop diagram (CLD) to represent the relationships in the SKOS example in Fig. 12. INSERT FIGURE 13. CAUSAL LOOP DIAGRAM According to the figure, poor-quality cow feed affects milk production, and in turn the processing time and final price of milk. This can be directly inferred from SKOS without constructing a CLD diagram by using the related, narrower, and broader SKOS properties. Likewise, considering how SKOS works to infer knowledge in a particular domain, a suitable model for a supply chain in SKOS can be provided. However, SKOSCM and the case study work only for the milk supply chain, although they may be applied to other supply chains by developing different models. Finally, Fig. 14 illustrates the data transformation process, first into SKOS and then into knowledge. INSERT FIGURE 14. TRANSFORMATIONS: RAW DATA TO SKOS AND SKOS TO KNOWLEDGE In Fig. 14, all supply chain echelons store raw data, such as purchase orders, inventory entries, and processing status, among other information concerning the milk supply chain workflow. This workflow consists of eight stages, as we noted earlier. To transform raw data into knowledge, SKOSCM stores the former in a SQL-based database, and it analyzes them to transform them into valuable data by using the SKOS representation provided by the milk supply chain model. For instance, if the price of alfalfa in a purchase order (for cow feed) is cheaper than in the last purchase, SKOSCM stores this variation as a “Change of Data.” In fact, the skos:broader relationship between alfalfa and cows is directly bound to milk production, since feed quality determines the amount of milk that cows produce and its and nutrients.

15 These stored data can then be transformed into knowledge through a SKOS engine. The engine stores changes, new data, and recurrent data (historical data) for variability detection and transform them into business information. In our case study, all this business information was transformed into knowledge when the production manager saw a decrease in milk derivatives production at the production stage, and performed a “Track Production” action to identify any issues. Then, the production manager saw that changes occurred in supplier selection, which is why the new quality of the alfalfa reduces the quality of the milk. Fig. 15 depicts this complete process of data transformation into knowledge. INSERT FIGURE 15. DATA-KNOWLEDGE TRANSFORMATION 4. EVALUATION Evaluating software systems is not an easy task, although different types of evaluations can be found in the literature. For instance, Kitchenham et al. (1997) proposed several quantitative, qualitative, and hybrid evaluation methods to evaluate software and tools, but these methods cannot always be useful for knowledge management system evaluation. An alternativ is quantitative evaluations based on the assumption that the software product has at least a measurable property that is expected to change as a result of using the methods/tools to be evaluated. Similarly, they can be developed in three different ways: case studies, formal experiments, and surveys. SKOSCM has measurable properties such as time for issue detection and precision in supplier selection. However, due to the nature of the data workflow among the supply chain echelons, it is hard to evaluate precision with SKOS, since this property is improved by the continuous participation of all supply chain echelons. Full participation can be achieved by using historical data for the analysis or to infer results and changes in supply chain behavior. Changes may be related to production, suppliers, and supply times, among others. As regards time, this property depends on how often SKOSCM is used in a complete workflow among all supply chain echelons. For this reason, it is only possible to measure the time needed for issue detection with and without SKOSCM. It is impossible to analyze other tools, since the supply chain of this

16 study has not employed any other tool similar to SKOSCM. Apart from the quantitative methods, the present research describes the term feature analysis as a qualitative evaluation. Feature analysis is based on both identifying the requirements that users have for a particular task or activity and mapping these requirements to features that a method/tool aimed at supporting that task or activity should possess. In this sense, ouir research relies on feature analysis to evaluate SKOSCM and measure its main aspects. 4.1. AHP Evaluation of SKOSCM As a feature analysis method, AHP is a decision-making process that can help people set priorities and make the best decision when both qualitative and quantitative aspects of a decision need to be considered. Saaty (2005) views AHP as a theory for the relative measurement of intangible criteria. It uses paired comparisons, unlike in traditional measurements where a scale is applied to measure elements, and they are measured one by one, not by comparing them one with another. Also, AHP has been used in the software engineering and IS literature (Thompson et al. 2001, Wieringa et al. 2006). It also is useful to evaluate processes (Vidal et al. 2011) and select product features (Subramaniana and Ramanathan 2012). In this case study, we used AHP to evaluate SKOSCM as a KMS. According to Ngai and Chan (2005), some of the main features for a KMS are: functionality (FU), document management (DC), collaboration (CL), communication (CM), and measurement (MA). However, in addition to these features, an expert panel2 selected for this study suggested that a KMS must have the following features: visual data representation (VS), data workflow management (DT), learnability (LR), and error detection (ED). 4.2. Evaluation Design The AHP evaluation was carried out in two steps. First, assessment features are selected; second, the weighted matrix is established according to these selected features. In this research, the second stage classifies the features that should be considered in a supply chain KMS. This assessment is also used to confirm that the features are used in SKOSCM in the case study. The AHP approach thus executes a pair The expert panel was formed with twelve experts: one knowledge management consultant, two warehouse operation managers, two transportation managers, one inventory control manager, one customer service manager, one logistics engineer, one purchasing manager, one production manager, one materials manager, and one vendor managed inventory coordinator. 2

17 comparison among selected features. A scale of nine points is used to execute the comparison, where 1 means ‘‘equally important,’’ 3 means ‘‘slightly important,’’ 5 means ‘‘more important,’’ 7 means ‘‘considerably more important,’’ 9 means ‘‘extremely important,’’ and the intermediate values of decision 2, 4, 6, and 8 mean "intermediate values." INSERT TABLE 6. WEIGHTED MATRIX OF ESSENTIAL KMS FEATURES Once the expert panel compared all of the features, the priority rank of each item was calculated. This stage is known as AHP synthesis, which starts by adding the values of each matrix column, and then each item is divided by the total of its column. The resulting matrix is called normalized pair comparisons matrix, as Table 7 shows. INSERT TABLE 7. NORMALIZED VALUES Then we can derive the general priority matrix, which shows the percentages obtained for each KMS feature, as shown in Fig. 16 (left). Finally, the five most important features of KMS in a supply chain context are obtained as shown in Fig. 16 (right). INSERT FIGURE 16. RELATIVE IMPORTANCE OF FEATURES According to Fig. 16, there are five most important features that any KMS application should possess: are: error detection (ED), data workflow management (DT), visual data representation (VS), measurement (MA) and documentation (DC). To provide a point of comparison, this evaluation was performed with eight other knowledge management systems: iLEAN, SAP, JDA Software, EPICOR, KINAXIS, UNIT4, and Oracle. The systems were selected based on their characteristics discussed earlier. The aim of this evaluation is to demonstrate that SKOSCM possesses the basic qualities of a KMS, such as functionality, collaboration, and communication, to mention a few. These features are in Table 8. INSERT TABLE 8. ESSENTIAL FEATURES OF A KMS IN THE SUPPLY CHAIN DOMAIN Evaluation Results. Table 9 presents the results of the compared KMSs and features.

INSERT TABLE 9. EVALUATION RESULTS This shows that SKOSCM and the compared systems contain most of the selected elements. However,

18 error detection (ED) is only present in SKOSCM, since the system requires a knowledge inference engine, while most other software solutions prefer exploiting data with data mining techniques and big data. Similarly, data workflow management (DT) is not present in any system, because it is a feature focused on tracking processes and detecting variations. It requires a large process of data acquisition to work. Finally, visual data representation (VS) is present in different ways ,such as charts for data interpretation, graphics for analytics, and visual data reports. In the case of SKOSCM, it provides a data workflow diagram (DWD) that represents lacks of data among echelons of the supply chain. Note that measurement (MA) and documentation management (DC) are not discussed; they are not relevant to SKOSCM. 4.3. Discussion The AHP evaluation demonstrates that SKOSCM is suitable for knowledge management. According to the expert panel, it also meets the needs of a supply chain for issue detection at the procurement stage. The two most important findings are: (1) the before and after performance of supplier selection in a supply chain, and (2) the opportunity to improve SKOSCM in order to detect live problems in a supply chain. As regards the performance of the company that implemented SKOSCM, Fig. 17 shows the time needed in hours to detect an issue with and without the software architecture. SKOSCM allows for timely issue detection of repetitive problems in the supply chain. As a result, low production peaks due to poor quality cow feed were reduced, since the inventory control manager could increase, cancel, or change the bill orders to fulfil the needs of the production plant. INSERT FIGURE 17. TIME USED FOR ISSUE DETECTION On the second finding, the SKOS approach takes advantage of previous data workflows as experience or historical data to detect issues in the early stages: issue detection in a “cold-start,” for example. And, since supply chains are domain-specific, a general solution is quite complex. Thus, domain-specific solutions are feasible. The modeled data workflow used in this case study can be used in other companies of the same domain – a milk supply chain –to solve a cold start. Additionally, once an issue is detected, if it can be reproduced, the data workflow works like a template for issue detection.

19 5. CONCLUSIONS AND FUTURE DIRECTIONS This article makes two major contributions. First, it shows how supply chain knowledge management can be improved at the procurement and production stages with semantic Web technologies such as SKOS. Second, it determines and discusses the most recommended features for a KMS. We also found that Web-based and semantic technologies provide platforms for the development of powerful applications but also opportunities of alleviating linguistic barriers to supply chain data across partners. Since knowledge management fails when supply chain partners do not communicate properly, SKOSCM detects issues in the transfer of a partner’s data and creates knowledge for the decision-making process. As far as future directions are concerned, this research faces three limitations. First, our case study addressed a particular business domain only: a milk supply chain. So SKOSCM must be proved that it can work in other domains. SKOS needs a controlled vocabulary to work appropriately, and knowledge management and acquisition depend only on the e-procurement system. Second, SKOSCM needs historical data and similar cases to detect recurrent problems. Future improvement to the system can address how to enhance system capability to detect new problems at the live stages of production. Third, the Web-based platform is currently a limited prototype version. For a collective supply chain, the platform must include collaborative and simultaneous support of e-procurement transactions. Hence, a SKOSCM must evolve into a software-as-a-service (SaaS) cloud-based platform.

REFERENCES Alor-Hernández, G., Sánchez-Ramírez, C., Cortes-Robles, G., Rodríguez-González, A., García-Alcaráz, J.L., Cedillo-Campos, M.G. 2014. BroSemWeb: a brokerage service for e-procurement using semantic web technologies. Computers in Industry 65, 828-840. Anand, N., Yang, M., Van Duin, J.H.R., Tavasszy, L. 2012. GenCLOn: an ontology for city logistics. Expert Systems with Applications 39, 11944-11960. Avelar-Sosa, L., García-Alcaraz, J.L., Cedillo-Campos, M.G., Adarme-Jaimes, W. 2014. Effects of regional infrastructure and offered services in the supply chains performance: Case Ciudad Juarez. Dyna 81, 208-217. Bailey, K., Francis, M. 2008. Managing information flows for improved value chain performance. International Journal of Production Economics 111, 2-12. Baker, T., Bechhofer, S., Isaac, A., Miles, A., Schreiber, G., Summers, E. 2013. Key choices in the design of Simple Knowledge Organization System (SKOS). Web Semantics: Science, Services and Agents on the World Wide Web 20, 35-49.

20 Becker, M.C., Zirpoli, F. 2003. Organizing new product development: Knowledge hollowing-out and knowledge integration-the FIAT Auto case. International Journal of Operations and Production Management 23, 1033-1061. Börjeson, N., Gilek, M., Karlsson, M. 2014. Knowledge challenges for responsible supply chain management of chemicals in textiles as experienced by procuring organisations. Journal of Cleaner Production 107, 130-136. Carrer-Neto, W., Hernández-Alcaraz, M.L., Valencia-García, R., García-Sánchez, F. 2012. Social knowledge-based recommender system: application to the movies domain. Expert Systems with Applications 39, 10990-11000. Chandra, C., Tumanyan, A. 2007. Organization and problem ontology for supply chain information support system. Data and Knowledge Engineering 61, 263-280. Chu, S.C., Sung, Y. 2015. Using a consumer socialization framework to understand electronic word-ofmouth (eWOM) group membership among brand followers on Twitter. Electronic Commerce Research and Applications. Davenport, T.H. 1994. Saving IT's soul: human-centered information management. Harvard business review 72, 119-131. de Sousa, J.P. 2000. Distributed planning and control systems for the virtual enterprise: organizational requirements and development life-cycle. Journal of Intelligent Manufacturing 11, 253-270. Espinal, A.C., Montoya, R.A.G. 2009. Information Technologies in Supply Chain Management. Dyna 157, 39. Floridi, L. 2008. The Blackwell guide to the philosophy of computing and information. Wiley, New York, NY. Flyvbjerg, B. 2006. Five misunderstandings about case study research. Qualitative Inquiry 12, 219-245. García-Cáceres, R.G., Perdomo, A., Ortiz, O., Beltrán, P., López, K. 2014. Characterization of the supply and value chains of Colombian cocoa. Dyna 81, 30-40. Göksu, A., Kocamaz, U.E., Uyaroğlu, Y. 2015. Synchronization and control of chaos in supply chain management. Computers and Industrial Engineering 86, 107-115. Gonnet, S., Vegetti, M., Leone, H., Henning, G. 2007. SCOntology: a formal approach toward a unified and integrated view of the supply chain. Chapter 7 in M.M. Cunha, B.C. Cortes, G.D. Putnik (eds.), Adaptive Technologies and Business Integration: Social, Managerial and Organizational Dimensions, IGI Global, Hershey, PA, 137-158. Gruber, T. 1993. What is an ontology? Available at TomGruber.org. Grubic, T., Veza, I., Bilic, B. 2011. Integrating process and ontology to support supply chain modeling. International Journal of Computer Integrated Manufacturing 24, 847-863. Hadikusumo, B.H.W., Petchpong, S., Charoenngam, C. 2005. Construction material procurement using Internet-based agent system. Automation in Construction 14, 736-749. Halley, A., Nollet, J., Beaulieu, M., Roy, J., Bigras, Y. 2010. The impact of the supply chain on core competencies and knowledge management: directions for future research. International Journal of Technology Management 49, 297-313. Hassan, H., Tretiakov, A., Whiddett, D., Adon, I. 2014. Extent of e-procurement use in SMEs: a descriptive study. Procedia - Social and Behavioral Sciences 164, 264-270. Hay, D. 2013. Data Model Patterns: Conventions of Thought. Addison-Wesley, Boston, MA.

21 Hjaila, K., Laínez-Aguirre, J.M., Zamarripa, M., Puigjaner, L., Espuña, A. 2016. Optimal integration of third-parties in a coordinated supply chain management environment. Computers and Chemical Engineering 86, 48-61. Holtbrügge, D., Berg, N. 2004. Knowledge transfer in multinational corporations: evidence from German firms. Management International Review 44, 129-145. Huang, H., Kauffman, R.J., Xu, H., Zhao, L. 2011. Mechanism design for e-procurement auctions: on the efficacy of post-auction negotiation and quality effort incentives. Electronic Commerce Research and Applications 10, 650-672. Huang, H., Kauffman, R.J., Xu, H., Zhao, L. 2013. A hybrid mechanism for heterogeneous e-procurement involving a combinatorial auction and bargaining. Electronic Commerce Research and Applications 12, 181-194. Hult, G.T.M., Ketchen, D.J., Arrfelt, M. 2007. Strategic supply chain management: Improving performance through a culture of competitiveness and knowledge development. Strategic management journal 28, 1035-1052. Johnston, M., Whang, S. E-business and supply chain management: an overview and framework. Production and Operations Management 11, 413-423. Kitchenham, B., Linkman, S., Law, D., 1997. DESMET: a methodology for evaluating software engineering methods and tools. Computing and Control Engineering Journal 8, 120-126. Koumoutsos, G., Thramboulidis, K. 2009. A knowledge-based framework for complex, proactive and service-oriented e-negotiation systems. Electronic Commerce Research 9, 317-349. Lau, H.C.W., Ho, G.T.S., Zhao, Y., Chung, N.S.H. 2009. Development of a process mining system for supporting knowledge discovery in a supply chain network. International Journal of Production Economics 122, 176-187. Lee, C.K.M., Lau, H.C.W., Ho, G.T.S., Ho, W. 2009. Design and development of agent-based procurement system to enhance business intelligence. Expert Systems with Applications 36, 877-884. Lee, J., Goodwin, R. 2006. Ontology management for large-scale enterprise systems. Electronic Commerce Research and Applications 5, 2-15. Lummus, R.R., Vokurka, R.J. 1999. Defining supply chain management: a historical perspective and practical guidelines. Industrial Management and Data Systems 99, 11-17. Luu, D.T., Ng, S.T., Chen, S.E. 2003. A case-based procurement advisory system for construction. Advances in Engineering Software 34, 429-438. Madni, A.M., Lin, W., Madni, C.C. 2001. IDEONTM: An extensible ontology for designing, integrating, and managing collaborative distributed enterprises. Systems Engineering 4, 35-48. Matheus, C.J., Baclawski, K., Kokar, M.M., Letkowski, J.J. 2005. Using SWRL and OWL to capture domain knowledge for a situation awareness application applied to a supply logistics scenario. Rules and Rule Markup Languages for the Semantic Web, Springer, Heidelberg, Germany, pp. 130-144. Miles, A., Bechhofer, S. 2009. SKOS simple knowledge organization system reference. W3C recommendation 18 August, W3C. Munoz, E., Capón-García, E., Laínez-Aguirre, J.M., Espuna, A., Puigjaner, L. 2015. Supply chain planning and scheduling integration using Lagrangian decomposition in a knowledge management environment. Computers and Chemical Engineering 72, 52-67.

22 Mylan, J., Geels, F., Gee, S., McMeekin, A., Foster, C. 2014. Eco-innovation and retailers in milk, beef and bread chains: enriching environmental supply chain management with insights from innovation studies. Journal of Cleaner Production 107, 20-30. Ngai, E.W., Chan, E.W.C. 2005. Evaluation of knowledge management tools using AHP. Expert Systems with Applications 29, 889-899. Niemi, P., Huiskonen, J., Karkkainen, H. 2010. Supply chain development as a knowledge development task. International Journal of Networking and Virtual Organisations 7, 132-149. Niemi, P., Huiskonen, J., Kärkkäinen, H. 2009. Understanding the knowledge accumulation process: implications for the adoption of inventory management techniques. International Journal of Production Economics 118, 160-167. Peiris, K.D.A., Jung, J., Gallupe, R.B. 2015. Building and evaluating ESET: a tool for assessing the support given by an enterprise system to supply chain management. Decision Support Systems 77, 41-54. Peng, J., Quan, J., Zhang, G., Dubinsky, A.J. 2016. Mediation effect of business process and supply chain management capabilities on the impact of IT on firm performance: evidence from Chinese firms. International Journal of Information Management 36, 89-96. Piramuthu, S. 2005. Knowledge-based framework for automated dynamic supply chain configuration. European Journal of Operational Research 165, 219-230. Rao, C., Zhao, Y., Ma, S. 2012. Procurement decision making mechanism of divisible goods based on multi-attribute auction. Electronic Commerce Research and Applications 11, 397-406. Rita, P., Krapfel, R. 2015. Collaboration and competition in buyer-supplier relations: the role of information in supply chain and e-procurement impacted relationships, assessing the different roles of marketing theory and practice in the jaws of economic uncertainty. In Proceedings of hte Academy of Marketing Science Annual Conference, Springer, Heidelberg, Germany pp. 98-105. Ruiz-Martínez, J.M., Valencia-García, R., Martínez-Béjar, R., Hoffmann, A. 2012. BioOntoVerb: A top level ontology based framework to populate biomedical ontologies from texts. Knowledge-Based Systems 36, 68-80. Ryoo, S.Y., Kim, K.K. 2015. The impact of knowledge complementarities on supply chain performance through knowledge exchange. Expert Systems with Applications 42, 3029-3040. Saaty, T. L. (2005). Analytic Hierarchy Process. Encyclopedia of Biostatistics. Wiley, New York NY. Sakka, O., Millet, P.A., Botta-Genoulaz, V. 2011. An ontological approach for strategic alignment: a supply chain operations reference case study. International Journal of Computer Integrated Manufacturing 24, 1022-1037. Schandl, T., Blumauer, A. 2010. PoolParty: SKOS thesaurus management utilizing linked data. The Semantic Web: Research and Applications. Springer, Heidelberg, Germany, pp. 421-425. Scheuermann, A., Hoxha, J. 2012. Ontologies for intelligent provision of logistics services. The Seventh International Conference on Internet and Web Applications and Services, Stuttgart, Germany. Scheuermann, A., Leukel, J. 2014. Supply chain management ontology from an ontology engineering perspective. Computers in Industry 65, 913-923. Service, U.E.R. 2007. Farm to table: the dairy supply chain. USDA Economic Research Service, Washington, DC. Shaw, M., Clements, P. 2006. The golden age of software architecture. IEEE Software 23, 31-39.

23 Sivakumar, K., Roy, S. 2004. Knowledge redundancy in supply chains: a framework. Supply Chain Management: An International Journal 9, 241-249. Soliman, K.S., Janz, B.D., Raisinghani, M.S., Meade, L.L. 2005. Strategic decisions in supply-chain intelligence using knowledge management: an analytic network process framework. Supply Chain Management: An International Journal 10, 114-121. Subramanian, N., Ramanathan, R. 2012. A review of applications of analytic hierarchy process in operations management. International Journal of Production Economics 138, 215-241. Sun, S.X., Zhao, J., Wang, H. 2012. An agent based approach for exception handling in e-procurement management. Expert Systems with Applications 39, 1174-1182. Taylor, R.N., Medvidovic, N., Dashofy, E.M. 2009. Software architecture: foundations, theory, and practice. Wiley Publishing. Thomas, D.J., Griffin, P.M. 1996. Coordinated supply chain management. European Journal of Operational Research 94, 1-15. Thompson, S., Fueten, F., Bockus, D. 2001. Mineral identification using artificial neural networks and the rotating polarizer stage. Computers and Geosciences 27, 1081-1089. Trappey, A.J.C., Trappey, C.V. 2004. Global content management services for product providers and purchasers. Computers in Industry 53, 39-58. Vidal, L-A., Marle, F., Bocquet, J.C. 2011. Using a Delphi process and the Analytic Hierarchy Process (AHP) to evaluate the complexity of projects. Expert systems with applications 38, 5388-5405. Wang, Y., Wallace, S.W., Shen, B., Choi, T.M. 2015. Service supply chain management: A review of operational models. European Journal of Operational Research 247, 685-698. Wen, H., Gu, Q. 2014. The elements of supply chain management in new environmental era. Proceedings of the Seventh International Conference on Management Science and Engineering Management. Springer, Heidelberg, Germany, pp. 867-880. Wen, W. 2007. A knowledge-based intelligent electronic commerce system for selling agricultural products. Computers and Electronics in Agriculture 57, 33-46. Wieringa, R., Maiden, N., Mead, N., Rolland, C. 2006. Requirements engineering paper classification and evaluation criteria: a proposal and a discussion. Requirements Engineering 11, 102-107. Williams, P.L., Jr. 2008. Factors affecting the adoption of e-business: an automotive industry study. Unpublished Ph.D. thesis, Capella Uiversity. Wu, D.J. 2001. Software agents for knowledge management: coordination in multi-agent supply chains and auctions. Expert Systems with Applications 20, 51-64.

24 APPENDIX 1. ADDITIONAL DETAILS OF THE SKOSCM ARCHITECTURE COMPONENTS To support our earlier discussion, we present additional details of the SKOSCM architecture next. •

Data Layer. This layer stores SCM data. Additionally, it contains all configuration tables that enable to operate the modules and services offered by SKOSCM. This layer includes two key components, the business data and the ontology, which are the core of knowledge of the entire software architecture. Similarly, this layer also ensures data acquisition across virtual private networks and the Internet by connecting with ERPs, CRMs, social networks, and other systems.

•

Data Management Layer. This layer communicates with the Data Layer in order to obtain business data and their representation through the ontology mapper to convert data to RDF.

•

SKOS Data Manager. This layer is responsible for knowledge management through SKOS. In this layer SKOS-based data is generated and consumed.

•

Integration Layer (API). This layer facilitates the creation of new SKOS-based applications through a series of public interfaces, which provide easy access to a set of services provided by the architecture. Services compositions are presented and defined in this layer. More specifically, the Data Manager component provides interaction between raw data and the user interface.

•

Presentation Layer. In this layer, SKOSCM determines the best way to display business data by using XHTML when HTML5 is not supported. The Presentation Layer does not know what events are taking place inside inferior layers and how the services are provided. It merely uses them to show the end-user interface. In this work, a graphic user interface is not provided; however, SKOSCM has been designed in order to support a Web-based user interface.

In addition to these layers, other important SKOSCM components include: •

Business Data. As part of the Data Layer, in the Business Data repository (a database of structured or nonstructured data), this component stores raw data from information exchange across the supply chain. It uses a data crawler to retrieve raw data from several sources, such as ERPs, CRMs, social networks, and structured data from other systems. Social networks are incorporated, since they help improve customer satisfaction and count on their opinions. For instance, Chu and Sung (2015) relied on a consumer socialization framework to understand electronic word-of-mouth (eWOM) group membership among brand followers on Twitter. Thus, a social network could provide insights on user satisfaction.

•

Ontology. This component manages data representation and the structure of supply chain knowledge. Both IDEON and SCOntolgy have been used in this component. Similarly, all knowledge represented in this layer ensures the hierarchical composition of relationships that can be used by SKOS Data Generator to build SKOS-based data.

•

SKOS. In thesauri and other structured knowledge organization systems, concepts can be classified into semantically meaningful bundles. Arrays are used to group specializations of a concept that share a common feature. For instance, the concept “cars” might be specialized in a first group of “cars by engine” (“ V8”, “ V6”, etc.), and in a second group of “cars by function” (“transport,” “family,” “sports,” etc.). Therefore, SKOS classifies and manages the vocabulary used by the Data Manager to retrieve processed and managed knowledge for the Presentation Layer.

•

Data Management. This component manages the petitions performed by the GUI Layer, specifically the user request of data and knowledge.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure16

Figure17

Figure15

25 Research work (Koumoutsos & Thramboulidis, 2009) (Wen, 2007) (Sun et al., 2012) (Luu et al., 2003) (Hadikusumo et al., 2005) (Lee et al., 2009) (Rao et al., 2012) (Huang et al., 2013) (Huang et al., 2011) (Wu, 2001) (Becker & Zirpoli, 2003) (Holtbrügge & Berg, 2004) (Munoz et al., 2014) (Ryoo & Kim, 2015) (Börjeson, Gilek & Karlsson, 2014) (Mylan et al., 2014) (Göksu, Kocamaz & Uyaroglu, 2014) (Peiris, Jung and Gallupe (Peiris, Jung & Gallupe, 2015) (Hjaila et al., 2015), (Peng et al., 2015) (Wang et al., 2015) SKOSCM

SKOS

SWT

SCM

KMS

KS

ARK

PS

SS

SA

RBR

CBR

BI

NO

YES

NO

YES

YES

NO

NO

NO

YES

NO

NO

NO

NO NO NO

NO NO NO

NO NO NO

YES NO NO

YES NO NO

NO NO YES

NO YES YES

NO NO NO

NO YES NO

YES NO NO

NO NO YES

NO NO NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO

NO NO NO NO NO

NO NO NO NO NO

NO NO NO NO NO

NO YES YES YES YES

NO YES YES YES YES

NO NO NO NO NO

YES YES YES YES NO

NO NO NO NO NO

YES NO NO NO NO

NO NO NO NO NO

NO NO NO NO NO

NO NO NO NO NO

NO

NO

NO

YES

YES

NO

NO

NO

NO

NO

NO

NO

NO

NO

NO

YES

YES

NO

NO

NO

NO

NO

NO

NO

NO

YES

YES

YES

NO

NO

NO

NO

NO

NO

NO

NO

NO

NO

YES

NO

YES

NO

YES

NO

NO

NO

NO

NO

NO

NO

YES

NO

YES

YES

NO

NO

NO

NO

NO

NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO

NO

NO

YES

NO

YES

NO

YES

NO

NO

NO

NO

YES

NO

NO

YES

YES

YES

NO

NO

NO

NO

NO

NO

NO

NO NO NO YES

NO NO NO YES

YES YES YES YES

YES YES YES YES

YES YES YES YES

NO NO NO YES

NO NO NO YES

NO NO NO NO

NO NO NO NO

NO NO NO NO

NO NO NO NO

NO NO NO YES

Table 1. Ontologies for supply chain management

26 Research Work (Koumoutsos & Thramboulidis, 2009) (Wen, 2007) (Sun et al., 2012) (Luu et al., 2003) (Hadikusumo et al., 2005) (Lee et al., 2009) (Rao et al., 2012) (Huang et al., 2013) (Huang et al., 2011) (Wu, 2001) (Becker & Zirpoli, 2003) (Holtbrügge & Berg, 2004)

SKOS

SWT

SCM

KMS

KS

ARK

PS

SS

SA

RBR

CBR

BI

NO

YES

NO

YES

YES

NO

NO

NO

YES

NO

NO

NO

NO NO NO

NO NO NO

NO NO NO

YES NO NO

YES NO NO

NO NO YES

NO YES YES

NO NO NO

NO YES NO

YES NO NO

NO NO YES

NO NO NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO

NO NO NO NO NO

NO NO NO NO NO

NO NO NO NO NO

NO YES YES YES YES

NO YES YES YES YES

NO NO NO NO NO

YES YES YES YES NO

NO NO NO NO NO

YES NO NO NO NO

NO NO NO NO NO

NO NO NO NO NO

NO NO NO NO NO

NO

NO

NO

YES

YES

NO

NO

NO

NO

NO

NO

NO

NO

NO

NO

YES

YES

NO

NO

NO

NO

NO

NO

NO

Table 2. Literature review highlights

27 Commercial Tool

SKOS

SWT

SCM

KMS

KS

ARK

PS

SS

SA

RBR

CBR

BI

iLEAN SAP JDA Software Manhattan Associates Epicor Infor IBM GTNexus Descartes Systems Group Kewill Systemss HihgJump Sofware PTC IBS IFS Inspur Genersoft Kinaxis Logility Quintiq Unit4 Totvs Oracle

NO NO NO

NO NO NO

YES YES YES

NO YES YES

YES YES YES

NO NO NO

YES YES YES

NO NO NO

NO NO NO

NO NO NO

NO NO NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO YES YES YES

NO NO NO NO

NO NO NO NO

YES YES YES YES

YES YES YES YES

YES YES YES YES

NO NO NO NO

YES YES YES YES

NO NO NO NO

NO NO NO NO

NO NO NO NO

NO NO NO NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO

NO

YES

YES

YES

NO

YES

NO

NO

NO

NO

NO NO NO NO NO NO NO NO NO NO

NO NO NO NO NO NO NO NO NO NO

YES YES YES YES YES YES YES YES YES YES

YES YES YES YES YES YES YES YES YES YES

YES YES YES YES YES YES YES YES YES YES

NO NO NO NO NO NO NO NO NO NO

YES YES YES YES YES YES YES YES YES YES

NO NO NO NO NO NO NO NO NO NO

NO NO NO NO NO NO NO NO NO NO

NO NO NO NO NO NO NO NO NO NO

NO NO NO NO NO NO NO NO NO NO

Table 3. Commercial tools for knowledge management

YES YES YES YES YES YES YES N/A N/A N/A N/A N/A N/A N/A N/A N/A YES

28 SKOS Property skos:broader skos:narrower skos:related

Features It enables the representation of hierarchical links, such as the relationship between one genre and its more specific species, or, depending on interpretations, the relationship between one whole and its parts. It enables the representation of associative (non-hierarchical) links, such as the relationship between one type of event and a category of entities that typically participate in it. It allows for the representation between two categories where neither is more general or more specific. It enables the representation of associative (non-hierarchical) links, which can be also used to represent part-whole links that are not meant as hierarchical relationships.

Table 4. SKOS standard properties

29 1 2 3 4 5

ex:cows rdf:type skos:Concept; skos:prefLabel "cows"@en; skos:related ex:milkproduction. ex:milkproduction rdf:type skos:Concept; skos:prefLabel "milkproduction"@en.

Table 5: SKOS Relationship between a Cow and Milk Production

30 Feature/Value FU DC CL CM MA VS DT LR ED Total

FU

DC

CL

CM

MA

VS

DT

LR

ED

1

3.000

7.00

3.00

0.33

0.111

0.111

5.000

0.111

0.333

1

3.000

3.000

5.000

0.111

0.111

3.000

0.111

0.143

0.333

1

1.000

0.167

0.143

0.143

1.000

0.111

0.333

0.333

1.000

1

0.200

0.125

0.125

0.111

0.111

3.000

0.200

6.000

5.000

1

1.000

1.000

4.000

0.200

9.000

9.000

7.000

8.000

1.000

1

0.200

9.000

0.333

9.000

9.000

7.000

8.000

1.000

5.000

1

9.000

0.333

0.200

0.333

1.000

9.000

0.250

0.111

0.111

1

0.111

9.000

9.000

9.000

9.000

5.000

3.000

3.000

9.000

1

32.010

32.200

42.000

47.000

13.950

10.601

5.801

41.111

2.422

Table 6. Weighted matrix of essential KMS features

31 Feature/Value

FU

DC

CL

CM

MA

VS

DT

LR

ED

Total

FU DC CL CM MA VS DT LR ED Total

0.031

0.093

0.167

0.064

0.024

0.01

0.019

0.122

0.046

0.576 0.682 0.179 0.162 0.867 1.455 1.971 0.348 2.756

0.01

0.031

0.071

0.064

0.358

0.01

0.019

0.073

0.046

0.004

0.01

0.024

0.021

0.012

0.013

0.025

0.024

0.046

0.01

0.01

0.024

0.021

0.014

0.012

0.022

0.003

0.046

0.094

0.006

0.143

0.106

0.072

0.094

0.172

0.097

0.083

0.281

0.28

0.167

0.17

0.072

0.094

0.034

0.219

0.138

0.281

0.28

0.167

0.17

0.072

0.472

0.172

0.219

0.138

0.006

0.01

0.024

0.191

0.018

0.01

0.019

0.024

0.046

0.281 1

0.28 1

0.214 1

0.191 1

0.358 1

0.283 1

0.517 1

0.219 1

0.413 1

Table 7. Normalized values

9

32 Feature Functionality (FU)

Description The software meets the user’s need and requirements.

Document Management (DC) Collaboration (CL) Communication (CM)

Search and organize documents for knowledge exploitation. Collaboration among teamwork partners for problem solving. Help the user work in a coordinated way and capture all crossgenerated data. Keep records of activities and changes of knowledge The capability for transforming knowledge into a visual representation for analysis. Allows for the tracking of data according to a predefined process.

Measurement (MA) Visual Data Representation (VS) Data Workflow Management (DT) Learnability (LR) Error Detection (ED)

How easily users can learn how to use the system. The capability to alert issues and critical processes “on live” by using data in a cold-start or a preconditioned start.

Table 8. Essential features of a KMS in the supply chain domain

33 Feature/Syst em ED DT VS MA DC

SKOSCM

iLEAN

SAP

JDA Software

EPICOR

KINAXIS

UNIT4

ORACLE

1 1 1 1 1

0 0 0 0 1

0 1 1 0 1

0 0 0 0 1

0 0 0 1 1

0 0 0 1 1

0 0 0 0 1

0 1 1 0 1

Table 9. Evaluation results

34

Highlights •

We present a software architecture by using SKOS for Supply Chain Knowledge Management

•

We identify the most suitable ontologies for developing SKOS-based Supply Chain Management Systems

•

We present a Web-based tool for automatic knowledge analysis in Supply Chains