Building Expert Profiles Models Applying Semantic Web Technologies

Semantic Web (SW) is an emerging research field that has application in different domains such as e-government services, richly interlinked library systems, Web search engines, enterprise knowledge stores, and other. The term “Semantic Web” refers to the World Wide Web Consortium’s (W3C) vision of the Web of linked data (called also the Web of Data) as “...an extension of the current Web in which information is given a well-defined meaning, better enabling computers and people to work in cooperation” (Berners-Lee, Hendler, & Lassila, 2001). Since then, many specifications, guidelines, languages, and tools have been developed that facilitate software development, improve performance and create new business opportunities.


Introduction
Semantic Web (SW) is an emerging research field that has application in different domains such as e-government services, richly interlinked library systems, Web search engines, enterprise knowledge stores, and other.The term "Semantic Web" refers to the World Wide Web Consortium's (W3C) vision of the Web of linked data (called also the Web of Data) as "…an extension of the current Web in which information is given a well-defined meaning, better enabling computers and people to work in cooperation" (Berners-Lee, Hendler, & Lassila, 2001).Since then, many specifications, guidelines, languages, and tools have been developed that facilitate software development, improve performance and create new business opportunities.This Chapter investigates the existing standard models for representing Human Resource (HR) data and especially in the context of competence management and discusses the process of building and publishing ontology-based expert profiles models.The expert profiles can be used for deploying advanced HR management services for in-house purposes, as well as integration and search of experts in the Linked Open Data (LOD) cloud.During the past 3 years, the Linked Data paradigm, promoted by Tim Berners-Lee (2006) as a part of the Semantic Web roadmap, evolved from a practical research idea into a very promising candidate for addressing one of the biggest challenges in the area of intelligent information management: the exploitation of the Web as a platform for data and information integration in addition to document search (Auer & Lehmann, 2010;Janev & Vraneš, 2011b).According to the statistics from March 2012 (see http://stats.lod2.eu/), the LOD cloud contains over 1,7 billion triples, integrates over 400000 namespaces and registers over 1,7 million instances of type "person".
The Chapter is organized as follows.After introducing the main requirements for managing competence data in enterprises in Section 2, we point to existing and emerging HR approaches and data representation standards (Section 3).Then, follows an example process for building an ontology-based enterprise knowledge stores (Section 4) and two examples for searching enterprise knowledge stores with Semantic Web tools (Section 5).

HR management requirements from business and technical perspective
Human resource management (HRM) is one of the basic business processes that consists of a wide range of administrative, organizational and employee acquisition and development activities.Administrative activities include management of different employee records (personal data, qualifications, holidays, business trips) and legal procedures of hiring / dismissal, as well as payment processing.Organizational activities cover strategic issues of enterprise organizations, systematization of working places, planning of team work, team formation and development, etc. Employee acquisition and development activities are directed towards definition of requirements, and standards that employees have to fulfil prior to employment, planning of necessary resources, education and development of employees, and employee performance measurement.
Herein, we would like to discuss the requirement for a comprehensive knowledge model for competence management from business and technical perspective.

Competence management business requirements
Competence management (CM) is an important research object in the more general area of human resource management.The idea of "competency" into the HR literature was introduced by the Harvard's psychologist David McClelland in early seventies of the last century (McClelland, 1973) and since then, development and use of competency based approaches within the corporate environment has been rapid (Draganidis & Mentzas, 2006).

Competence management on company level
Companies adopt different competency models and start competence and skills management initiatives in order to create a setting for the empowerment of their workforce and thus increase competitive advantage, innovation, and effectiveness (Houtzagers, 1999).A competency model is a list of competencies which are derived from observing satisfactory or exceptional performance for a specific occupation or task.Related to in-house competence management mainly aimed at building individual competence models are the following requirements:


building central repositories which define competencies for certain communities;  building services for identifying experts and finding out and continually recording what people ("experts") in an organization know ("expertise");  making expertise available to users so they can answer questions or solve problems that exceed personal or workgroup capabilities;  expertise gap analysis;  planning the expertise development paths; etc.

Competence management for cooperation and integration of activities with partners on national and international level
In order to be competitive in the global knowledge economy, companies organize themselves in partner networks or even virtual enterprises that require interlinking of activities, or even existing information systems.Business processes in such networks often spawn different specific tasks that are to be solved by the network members.Therefore, it is essential that partner organizations prove themselves with complementary competencies both on an expert and an organizational level.Developing and maintaining competence profiles of all the relevant parties associated with specific task and topic can significantly improve the performance of the partner network or the virtual organization.Related to inter-enterprise cooperation, interoperability and integration are the following requirements:  standard description of occupations and competences ;  using multi-lingual dictionaries for building expert profiles;  interoperability of knowledge models with similar schemas on the Web; etc.

Technical requirements
In order to achieve transparency and comparability of expertise, organizations need tools and technologies to express the core competencies and talents of employees in a standardized, machine processable and understandable format.Based on the competence management business requirements briefly introduced above, we can distinguish three types of expertise management services, namely (1) expert profiling and search, (2) organization profiling and search and (3) knowledge items search and retrieval.
Technologies that play a role in implementation of these services originate from the fields of open systems architecture, Web services, information retrieval, data and text mining, clustering, natural language processing, ontology building, etc.

Analysis of classical approaches to expertise management
The actual HRM solutions mainly focus on the integration of the distributed legacy databases, typically in the form of the data warehouse where the fact data (i.e.employee data) is arranged in order to answer the analytical queries efficiently.Personal profiles here usually rely on the self declared expertise.Employees keep track of their areas of expertise manually by maintaining a list of keywords or phrases and this list of key qualifications is being defined in the HR sector.This approach is error-prone since users are typically subjective/biased and reluctant to update the file regularly.Also, manually created lists cannot be an exhaustive description of the person's expertise areas.In addition, content based approaches (Sim et al., 2006) to expertise extraction, profiling and finding have been introduced lately that focus on the automatical identification of the expertise entities in the semi-structured and unstructured documents containing the expertise information as well as on the annotation of the identified expertise entities with the semantic mark-up.The input documents are: (1) curricula vitae and résumé that have been published in formats such as text, PDF, DOC and HTML; (2) publications and other legacy documents (Balog et al., 2006;Balog & de Rijke, 2008); (3) e-mails, blog sites and other online social networking related context (Aleman-Meza et al., 2007;Schäfermeier & Paschke, 2011).The expertise extraction and profiling is based on the linguistic analysis, statistical and machine learning classification methods as well as on the inductive logic programming techniques to discover rules for extracting fields from documents (Fang & Xiang Zhai, 2007;Petkova & Bruce Croft, 2006;Jung et al., 2007).Inspired by different research fields such as expert finding, competency management, terminology extraction, keyword extraction and concept extraction (Bordea, 2010), Bordea and Buitelaar (2010) proposed a hybrid approach and the Saffron system for expert profiling and finding.

Review of standards and literature for ontology-based competence management
European Union, through its chief instruments for funding research (FP5 -The Fifth, FP6 -The Sixth and FP7 -The Seventh Framework Programs), has financed several projects that focused on ontology-based competency management.As a result of these projects, several prototype systems have been developed (Bizer et al., 2005;Draganidis et al., 2006) Schmidt & Kunzmann (2006) developed the Professional Learning Ontology that formalizes competencies as a bridge between human resource development, competence and knowledge management as well as technology-enhanced learning.In (Bizer et al., 2005), Bizer developed a HR ontology, an e-recruitment prototype and argued that using Semantic Web technologies in the domain of online recruitment could substantially increase market transparency, lower the transaction costs for employers, and change the business models of the intermediaries involved.In (Paquette, 2007), the author presented a competency ontology and the TELOS Software Framework for Competency Modelling and Management.
Furthermore, the research work in the competence management domain in the last decade had a positive impact on several European Public Employment Services, e.g.see DISCO project (Müller-Riedlhuber, 2009).Some of them have already introduced (e.g.Germany, Norway) or are at the moment working on improvements of their matching (vacancies and job seekers) processes by shifting more emphasis to competences.

Explicit representation of an HR knowledge store
To represent information on the Web and to ensure interoperability between applications that exchange machine-understandable information, the Semantic Web uses the Resource Description Framework (RDF) as a general-purpose language.RDF describes information in terms of objects ("resources") and the relations between them via the RDF Schema, which serves as a meta-language or vocabulary to define properties and classes of RDF resources.The next layer on top of the RDF/RDFS data model serves to formally define domain models as shared conceptualizations, also often called ontologies (Gruber, 1993).
Ontologies are nowadays very often used for building integrated inter-and intraorganization business services, and to make the search and retrieval both efficient and meaningful.In this Section we will use the RDF and OWL languages to introduce the most important concepts and relations between concepts relevant for building an expert profile (see Figure 2).In a top-down manner, we can define the most general concepts/properties as subclasses / subproperties of entities from the public vocabularies FOAF, DOAC, and BibTeX (Aleman-Meza et al., 2007)and assign them meaning identical with the existing commonly used classes in the Semantic Web (see Table 2).In that way, the main "components" are defined as subclasses of the public concepts (foaf:

From implicit to explicit data representation
After the ontological knowledge base is designed, the next step is to populate the ontology i.e. import data into the ontology and create instances.Manually creating of ontologies is a time consuming task.Semantic Web community has delivered many high-quality opensource tools that can be used for automatic or semiautomatic ontology population i.e. to convert the facts trapped in the legacy systems or business documents into information understandable both for machines and people.
Professional HRM systems, e.g. the SAP Human Capital Management solution, cover the whole life-circle of an employee from her/his recruitment, training, development, and deployment to retirement.They enable tracking of employee movements and adequate tracking of changes in organizational structure.Furthermore, standard SAP HCM processes support skill management and give managers and HR professionals reporting and analysis options that provide a real-time insight into employee qualifications.As a result, the underlying (implicit) data base model is highly normalized and quite complex.Customizing the predefined SAP HCM functionalities or extending them with new client tailored functionalities require SAP consultancy efforts.Therefore, extracting the HR data in explicit format and enriching them with semantic information will make the data easily accessable and processable in other business applications.Table 3 gives an example how specific groups of data, called "infotypes" in SAP terminology, can be mapped to public or in-house defined domain classes.

Publishing and searching enterprise knowledge stores with SW tools
Once represented as an RDF data store, the HR data can be linked in the (LOD) cloud and become available for further exploitation.Herein, we would like to discuss two possibilities for searching RDF models: 1. using OntoWiki tool (Auer, 2007); 2. using Sig.ma, a service and an end user application to access the Web of Data (Tummarello et al., 2010).

Navigating and querying the semantic knowledge models with OntoWiki
In order to publish the developed ontological models on the Web, maintain, search and retrieval in efficient and meaningful way the OntoWiki Knowledge Engineering opensource tool can be used (see ontowiki.net).The main goal of the OntoWiki is to facilitate the visual presentation of a knowledge base as an information map, with different views on instance data.OntoWiki provides a generic user interface for arbitrary RDF knowledge bases.Each node at the information map, e.g.RDF resource foaf:Person, is represented as a Web accessible page and interlinked to related digital resources, e.g. using the rdfs:subClassOf semantic property to other RDF resource foaf:Agent.
Selection opportunities include (see Fig. Once a selection is made, the main content section will arrange matching content in a list view linking to individual views for individual instances.The right sidebar offers tools and complementary information specific to the selected content.
The main steps in the process of navigating and querying of a semantic model can be summarized as follows.
1. Select a knowledge base e.g.Organization and Personal profiles; 2. Select a semantic concept e.g.foaf:Agent; 3. Filter the entities using a semantic relation e.g.rdf:type in order to retrieve all instances of type foaf:Person; 4. Filter the entities with the faceted navigation filter e.g.retrieve the personal data for a person with a surname Janev; 5.After reviewing the results, the user may wish to continue navigating the information space by following relations between instances e.g.foaf:PrimaryTopic -1 can be selected to link the instance Janev with its personal profile document 1526-PPD (see Fig. 5).Links to the MPI document base that stores the publications and other documents created in the MPI working process are framed in red (Janev et al., 2010).

Searching Web of Data using Sig.Ma
Once available on the Web, expert profiles can be searched with Semantic search engines, e.g.Sig.Ma.
Fig. 6.An example of available personal profile document in the LOD cloud.

Conclusion
Taking into account the new trends in the design and implementation of enterprise information systems (based on adaptable, flexible, and open IT architecture, using open standards and emerging technologies), this Chapter introduced new insight into expertise management and proposed the Semantic Web-based approach to HR data representation, integration and retrieval.
Ontology-based approach to competency management: The proposed ontology-based approach to competency management includes establishment of a modular knowledge base of expert profiles and population of the knowledge base with information extracted from different HR related sources.The proposed approach based on emerging technologies and tools does not complement the existing information -integration approach (e.g.integrating the expert data in a form of a database) or the content management approach (e.g.integrating the experts' documents in a form of a document base), but it rather extends, enhances and integrates them with the aim to obtain a complete picture of the available resources.
Explicit, standard format of expert profile that facilitates data interoperability and expert search: As the interoperability between different knowledge organization schemas is one of the major Linked Open Data issues, the design of the semantic knowledge model in this Chapter was based on public vocabularies such as FOAF, DOAC, SIOC, DOAC, BibTeX, as well as common vocabularies for modelling case study specific data and relations such as DC, RDF, RDFS, and OWL.
Enhancing self-declared expertise with competences automatically and objectively extracted using text analysis: Taking into consideration that self declared expertise cannot be an exhaustive description of the person's expertise areas, the use of text analysis tools for updating the semantic expert profiles with uncovered latent knowledge should be considered.
Meaningful search and retrieval of expertise: Recently, a new search approach has emerged.It has been named faceted search that combines the navigational search paradigm and the direct, keyword search paradigm.Faceted search methods augment and improve traditional search results by using not just words, but concepts and logical relationships that are components of an ontology.Faceted navigation techniques and semantic relations shorten the search time, improve the relevance of search results, and deliver high-quality search services.This Chapter demonstrates the use of these methods in practice.

Acknowledgement
This work was supported by a grant from the European Union's 7th Framework Programme (2007)(2008)(2009)(2010)(2011)(2012)(2013) provided for the project LOD2 (GA no.257943), as well as by the Ministry of Education and Science of the Republic of Serbia under grant agreement no TR-32010 (SOFIA -Software Framework for Intelligent Adaptive Management of Complex Facilities).
HR-XML Consortium Competencies Schema, http://ns.hr-xml.org/SOC The 2010 Standard Occupational Classification (SOC, www.bls.gov/soc)system is used by Federal statistical agencies to classify workers into occupational categories for the purpose of collecting, calculating, or disseminating data.All workers are classified into one of 840 detailed occupations according to their occupational definition.To facilitate classification, detailed occupations are combined to form 461 broad occupations, 97 minor groups, and 23 major groups.O*NET The Occupational Information Network (O*NET, http://www.onetcenter.org,based on SOC) is designed to be the nation's most comprehensive resource of occupational information, with a database system that includes 275 descriptors about each occupation.DISCO European Dictionary of Skills and Competencies, financed by EU Leonardo da Vinci programme & the Austrian Federal Ministry for Education, the Arts and Culture, http://www.skills-translator.net/e-CF European e-Competence Framework, a reference framework of 32 ICT competences, http://www.ecompetences.eu/.ESCO The European Skills, Competences and Occupations taxonomy (under development).A partial classification is already in use in the European job mobility portal EURES (http://ec.europa.eu/eures/).It exists in 22 languages and currently contains around 6000 skill descriptions and 5000 job titles.

Table 1 .
International HR initiatives

Table 2 .
RDF modelsFor example, the Mihajlo Pupin Institute ontology (MPI) uses concepts from the DOAC+FOAF vocabulary and extends them with new concepts and properties defined in the imp and skills namespace as follows: SIOCThe SIOC initiative (Semantically-Interlinked Online Communities, http://www.w3.org/Submission/2007/02/) aims to enable the integration of the online community information.SIOC provides the Semantic Web ontology for representing rich data from the Social Web in RDF.FOAFThe FOAF (Friend of a Friend, http://www.foaf-project.org/)project is about creating a Web of machine-readable pages describing people, the links between them and the things they create and do.DOAC DOAC (Description Of A Career, DOAC Vocabulary specification, http://ramonantonio.net/doac/0.1/) is a vocabulary used for describing professional capabilities of a worker.It was designed to be compatible with the Europeans Curriculum so that those can be generated from a FOAF+DOAC file.DOAP DOAP (Description of a Project, DOAP Vocabulary specification (http://trac.usefulinc.com/doap/ ) is a RDF schema and XML vocabulary used for describing software projects and, in particular, open source.Dublin CoreThe Dublin Core Metadata Initiative (http://dublincore.org/) is an open organization engaged in the development of interoperable online metadata standards that support a broad range of purposes and business models. relations between a person and his/her expertise (imp:degree, imp:graduationTitle, imp:useDBMS, imp:useModellingTool, imp:useProgrammingLanguage);  relations between a person and the document base (foaf:workInfoHomepage, foaf:workplaceHomepage); etc.

Table 3 .
Establishing correspondence between implicit and explicit data representation