Semantic Web Approach for Determining Industry Qualifications Demand on Real-time Bases S. AbdElall, C. Reise, G. Seliger Technische Universität Berlin, Institute of Machine Tools and Factory Management, Pascalstr. 8-9, 10587 Berlin, Germany
[email protected] Conference Topic: University-business cooperation Keywords: Qualifications, curriculum, job announcements, semantic web, ontology
INTRODUCTION Prior applying for a job, candidates have to choose courses and curriculum that will qualify them for their intended job position. In times of dynamised markets, qualification demands change more frequently. Identifying the industry qualifications demand on real-time bases becomes therefore an important issue. Job announcements are written carefully to reflect exactly the current qualifications demand whenever a job vacancy appears. Qualifications described in job announcements vary often in their verbal descriptions (syntactics) even though the specific requirements (semantics) may be the same. For a meaningful analysis it is useful to decrease the amount of categories by bringing synonyms together under one qualification description. An automated analysis of job announcements has the potential to be more resource effective and more quickly than a human analysis. Modern information technology offers the chance to identify same semantic by using artificial intelligence. It is based on a model describing the classification of qualification needs. The information technological model – a so-called ontology – aims to be a formal and explicit description of a shared concept about technological qualification. It brings different perspectives on things together by defining its relations and memberships to a class. Therewith machines are enabled to reason. They can transform existing, but implicit, knowledge into explicit one, using socalled first order logic. Such logics using ontologies – the so-called sematic web – enable software applications to interpret heterogeneous information. Semantic Web knowledge bases facilitate precise and unambiguous answers in web queries. The volume and diversity of information can be used more efficiently, and knowledge can be better distributed due to the creation of a vast global knowledge base. New information can be gained through reasoning and used in knowledge intensive applications, such as analyses of job announcements regarding their qualification demands. In this paper, more than one thousand job announcements have been analyzed to determine the real-time job market qualifications demand for the manufacturing engineering. For a precise analysis, custom software has been programmed, this software is cumulatively learning by analyzing each job announcements, the ontology and its application on a big database with many entities
representing qualification statements is presented. The ontology was modeled with the open source software protégé, the database is created using a Microsoft Access databases. 1 1.1
ENGINEERING EDUCATION Education system
The education system can be viewed as a production system. The student entered the system as an input, shaped by several teaching activities, at the end of study period student graduated as an output of the system. Education can be considered as a process that transforms a student into a graduate. The input and output quantities are attributes of the student and can be described in terms of understanding, knowledge, skills, abilities etc. The process itself can be described as the sum of all activities of the educational institute, such as giving lectures and organizing laboratory experiments [1]. The graduated student as an output of the education system should satisfy the demand of the customer. The customer in this case is the industry that is intended to employ the graduate. The educational institution as a supplier of graduate must be aware of the customer qualifications demand, therefore producing a qualified graduate according to these demands. This interaction between educational institutions and industries can be viewed as a chain of suppliers and customers [2]. 1.2
Engineering curriculum
Engineering curriculum is mostly divided into three parts: compulsory courses, elective courses, and senior project. Each counted for 45%, 30% and 25% of the total European Credit Transfer and Accumulation System (ECTS) respectively [3]. The higher education institutions push the graduates with certain qualifications through the chain to satisfy the industry demand. The push system results in high inventory level and the pull system is difficult to implement when the lead-time is too long as impractical to react to demand information [4, 5]. Hybrid strategy of combining the two, Push-Pull Strategy at the higher education will reduce the gap between the qualifications supply and demand as shown in Fig. 1. The compulsory courses are at the initial stages of the engineering study and it follows the push strategy. These courses contain the fundamental or basic subjects e.g. science, mathematics, ethics, sustainable development etc. Latter at the senior level stage, the student must be aware of the industry-oriented qualifications demand. Hence, choose the courses that will qualify them for their intended job position, this follows the pull strategy i.e. courses should be selected according to recent job market demand. Further, senior project is a good opportunity for students to be more specialized in specific area of work and to enhance the communication and teamwork skills.
Fig. 1. Push/pull concept in engineering curriculum
2 2.1
REAL-TIME ENGINEERING QUALIFICATIONS Engineering qualifications
Qualification can be refers to the capacity, knowledge, or skill that makes someone suitable for a particular job or activity [6]. The strength of a person in possessing these qualifications, determine his performance quality. The success of educational institutions in supplying the industries with graduates who possess the required qualifications depends mainly on the level of understanding the job market qualifications need. Several studies and researches are annually conducted to determine the job market qualifications demand, all of which claim to have been produced in consultation with employers [7-13]. The methodology used in these studies are questionnaires, interviews, and/or brainstorming with the employers. The result of such research is dynamic due to technological, governmental and social changes, which mean results are subjected to continuous changes. Furthermore, finding the specific or detailed qualifications for each job title is hard to achieve, and if achieved the results will change in the near future. This makes these approaches unreliable in long-term perspectives. 2.2 Concept In any industry whenever there is a job vacancy or a need for specific skills, they translate these needs into a form of job announcement. Job announcement contains valuable information of the assigned responsibilities, and qualifications needed for specific job position. The main part of the job announcement is the qualifications demand; if the applicants satisfy these demands, they will have a high chance of getting the job. Job announcements can be considered as the voice of the industries, which reflect their current qualifications demand. Industries are continuously updating these data according to their realtime demands. The analysis of job announcements will provide valuable information about the current qualifications demand. Higher education institutions can use these data for shaping/updating their curricula as discussed in section 1.2. Training/teaching individual based on the latest job market demand will help to better-fit individuals to the industry demand i.e. the qualifications supply by educational institutions will be according to the industry qualifications demand. This concept model for preparing individual based on realtime industry qualifications demand is shown in Fig. 2.
Fig. 2. Real-time engineering qualifications
This model can help for improving the efficiency of information sharing between industry and educational institutions. Higher education institutions will be provided with an efficient communication channels with the employers, hence update the education programs that matches the job market demands. Teachers will be continuously aware of these changes in their area and consider how these can be translated into learning outcomes. These learning outcomes will be offered for the students in the shape of course/courses which they can apply for. Students will be aware of the job market qualifications demand and hence select the courses to develop themselves with the most relevant qualifications that will equip them in the best way for a rewarding career of their choice. Semantic web technology can help in creating such intelligent system, hence reducing the gap between the job market and the educational institutions by preparing work-ready graduates. This system will be able to collect and analyse the job market qualifications demand in real-time, and proposing these demands in term of qualifications to the educational institutions for preparing the courses and curriculum. 3 3.1
SEMANTIC WEB TECHNOLOGY How to cope with too many information on the web?
The World Wide Web (WWW) offers a vast potential to provide vast amounts of information globally. The continuous growth of job announcements published on the WWW gives job seekers the chance to find jobs directly by surfing the web. Humans understand the semantics of the published texts in the job announcements in term of qualifications needed, and responsibilities assigned for the stated job position. The Semantic Web is an extension of the existing WWW. It intends to bring structure to the meaningful content of Web pages. With additional data about data, so-called Meta-Data it creates an environment where software agents are able to roam from page to page and carry out sophisticated tasks for users [14]. Meta-Data describes documents or pictures and enables machines to understand the provided information in order to find and use information. To guarantee the standardized description of the contained data, ontologies are used. An ontology is “a formal model of an area of application that is used to facilitate the exchange and the distribution of knowledge” [15]. Ontologies represent the structure of the knowledge base in a defined model [16]. Entities of knowledge are stored based on the ontology structure in a so-called semantic web knowledge base. The main task of the knowledge base is to store data about a certain topic. In this way, the data can be found and combined from independent software agents for various purposes, which can access the knowledge base via internet, e.g. job search assistances [17]. 3.2
Existing semantic web software tools
The use of software tools has increased during the last years drastically in all kinds of application processes. Nowadays, there are many specialized tools that are used from companies, head hunters and job seekers to publish, search, compare and analyse job announcements. In Germany online job markets/searching engines/machines have already replace the traditional press media as the main source of information about job announcements. 66 % of the job-seekers prefer using online job markets/searching engines/machines [18]. Widely used job portals between companies, head hunters and applicants are for instance Step Stone, Experteer, Jobscout24 and Monster. Next to these classic job markets which are all using highly developed searching algorithms, semantic search engines are advancing on the market. JobiJoba.com is a project of the French IT-start-up ALLGOOB and is one of the leading job search engines in France. They recently opened up international offices in UK, Canada, USA, Spain, Belgium and Germany, where they are since 2011 with their sematic search online supporting some of the big job platforms [19]. At the moment they offer about 200,000 job announcements which are enriched with meta data for improved search results. StepStone,
one of the leading German online job portals, has acquired Jobanova GmbH, a specialist for semantic search technology [20]. Monster purchased Trovix, who is specialised on semantic search [21]. “Jobsucher.ch” is the first semantic job search engine and directly searches the 80,000 vacancies on company webpages in Switzerland. The research project “SABINE„ researched in Germany how semantic technologies can be implanted into application management. Job-seekers can upload their resumes into the system and get supported by a competence-based, semantically-adapted job search [22]. 3.3
Implementation
The ontology model is based on a vocabulary empirically developed out by analyzing around 1100 job announcements for job titled manufacturing engineer. The data have been collected from the period 01.02.2011 to 30.04.2011. The sources of these data were several job search engine like Monster (www.monster.com), Jobsearch (www.jobsearch.co.uk), and Bayt (www.bayt.com). A customized software application has been programed for analyzing the job announcements and determining the job market qualifications demand. The qualifications and its defined vocabulary were formalized as ontology and enriched with a knowledge base, with various relations between qualification statements. The ontology allows the interoperability between web sites e.g. company career forums, databases e.g. with job announcements, and software systems e.g. search engines. The model enables the storage of instances consisting of very specific as well as general qualifications statements. The adaptability of ontologies makes the knowledge base expandable and modifiable, and this, in particular, conforms to the dynamic requirements of the job market in a very successful way. For the information technological representation of the model the software tool Protégé was used. “The software tool Protégé is a free, open-source platform that provides a user with a suite of tools to construct domain models and knowledge-based applications with ontologies”. It implements a set of knowledge-modeling structures and actions that support the creation, visualization, and manipulation of ontologies in various representation formats. Further, Protégé can be extended by way of a plug-in architecture and a Java-based Application Programming Interface (API) for building knowledge-based tools and applications [23]. Fig. 3 shows a Screenshot of the classification of knowledge and capabilities in Protégé. The analysis of the job announcements brought up a catalogue of 3371 records, each record contains a set of keywords and the synonym qualification. Each qualification could be presented by several records, so the total number of the unique qualifications was 1518. This catalogue formed the dictionary base for creating the qualification ontology. But the catalogue only includes highly instantiated aspects of qualification, i.e. common qualification keywords that have often been used in announcements. These qualification classifications are mostly located on the lowest level of qualification hierarchies, e.g. specific knowledge about production technologies like machinable materials or mastering various manufacturing techniques. The characteristics of terms can be classified by many aspects, e.g. materials can be classified by their mesh-texture, hardness or physical and chemical composition in relation to the manufacturing process they are used in. Proficiency in manufacturing technologies on the other hand can either be specified by the required process steps or by the DIN/ISO guidelines. Based on the qualification description there could be both 1:n and n:1 relations between qualifications which can widen or narrow the hierarchical order. The first step was to define categories that represent the qualification spectrum of a manufacturing engineer. In the model the qualifications are divided in two main fields, Knowledge and Personal Capabilities. The Knowledge domain comprises engineering, general sciences, IT and management. Engineering knowledge comprises methods and engineering related expertise, both important competencies for a manufacturing engineer, e.g. manufacturing techniques, lean production, etc. General Sciences represent technical and methodical knowledge fields, which are not
directly connected to engineering. Also important for the modern engineer is the knowledge about management processes such as project management, finances, markets, etc. Supporting IT skills are for example programming, database or MS office proficiency. Personal Capabilities include all personal and social aspects that characterize one person’s ability to Drive, Focus, Guide and Impact. This taxonomy is used by the Siemens Company in its competency model [24]. A person’s Drive is defined by his/her ability to take initiatives, creativity or decision making skills. Focus includes the person’s dedication to lifelong learning, analytical skills and organization. Capabilities like assertiveness, communication and networking skills are essentially to be able to assert innovation and vision. A modern manufacturing engineer interacts in a dynamic and interdisciplinary project environment in which team skills, motivation, inspiration and situational sensitivity are most important. Those characteristics are included within the category Guide. Fig. 4 shows the top classes and some selected subclasses of the model.
Fig. 3. Screenshot of the classification of knowledge and capabilities in Protégé
Fig. 4. Classification of knowledge and capabilities In the scientific area of Manufacturing Engineering and other scientific fields hierarchical orders have been taken of existing classifications, e.g. from scientific publications, norms and guidelines like CIRP and DIN/ISO. In addition hierarchies from teaching in manufacturing technology and factory management have been adopted. The next step after creating the hierarchical classification was to define and represent the relations between keywords. There are synonyms occurring like “continuous improvement process” and “kaizen”. For example the method “root cause analysis” is a sub-class of “problem solving skills” or “Catia” and “SolidEdge” are sub-classes of “CAD”. In the subclass “production planning” knowledge about the bill of materials and the work instructions are so-called cousins because they are both needed to know what has to be produced (job/work order) and how it will be done (production schedule). In addition to hierarchical relations there exist relations between classes in different subsystems. For instance you need for a flow simulation of the material in the shop floor knowledge about the analysis technique “Material Flow Analysis” which belongs to the “Lean or Factory Management” knowledge. You also need the ability to use a “simulation tool”, to simulate different options and scenarios. Dependent on the model, these classes can be assigned to different hierarchies like “Lean or Factory Management” and “Simulation tool” which have to have a relation for modeling the fact that a successful “material flow planning” need skills and abilities of both.
Fig. 5. Relations and implications in a specific knowledge With the knowledge model it can be derived automatically that a person X, who has expert knowledge in Lean Production has to have at least basic knowledge about assembly lines. This result is derived out of the following logic (see also Fig. 5): To be an expert in lean production, person X has to have amongst others deep knowledge in buffer storages, onepiece-flow as well as U-Layouts for work places. But if person X has deep knowledge in the latter three disciplines it can be implied that he/she also has at least basic knowledge in the knowledge areas storage systems, material flow planning and workplace layout, because they are root categories of the named qualification statements belonging to Lean Production. Those disciplines are on the other hand defined subclasses of assembly lines. Thus the information that Person X has expert knowledge in lean production infers that he/she also has at least basic knowledge in assembly lines. 4
OUTLOOK: HOW TO MATCH QUALIFICATION OFFER A WITH DEMAND B
To create efficient knowledge based systems for mapping qualification profiles several questions have to be solved before they will provide ubiquitous services in managing qualification. In the area of metadata annotation there are several qualification models with different expressiveness in usage. Also, the annotation of qualification statements in web documents is not broadly accomplished. Additionally, the creation of relations and algorithms needs further research in order to visualize dependencies and taxonomic similarities of qualification profiles for engineers. Nevertheless the potentials of Semantic Web Technology show the possibilities to increase the management of qualifications tremendously through individual query and provisional knowledge. The idea of knowledge based systems working on a global qualification-database is within reach. As shown above first applications in the job market show promising beginnings. REFERENCES [1]
Rompelman, O. and E. De Graaff, The engineering of engineering education: curriculum development from a designer's point of view. European Journal of Engineering Education, 2006. 31(2): p. 215-226.
[2]
Al-Turki, U.M., et al., Stakeholders integration in higher education: supply chain approach. European Journal of Engineering Education, 2008. 33(2): p. 211-219.
[3]
GPE. Global Production Engineering - The master program of the TU-Berlin. 2012 [cited 2012 12.03]; Available from: www.gpe.tu-berlin.de.
[4]
Simchi-Levi, D., P. Kaminsky, and E. Simchi-Levi, Designing and managing the supply chain : concepts, strategies, and case studies. Irwin/McGraw-Hill series. Operations and decision sciences2000, Boston: Irwin/McGraw-Hill. xxii, 321 p.
[5]
Chopra, S., Supply chain management : strategy, planning and operation. 3rd ed. ed2007, Upper Saddle River, NJ: Pearson Prentice Hall ; [London] : Pearson Education. xv, 536 p.
[6]
Oxford-Dictionaries. qualification. 2011 [cited 2011 27.12.2011]; Available from: http://oxforddictionaries.com/definition/qualification?q=qualification.
[7]
Nguyen, D.Q., The essential skills and attributes of an engineer: a comparative study of academics, industry personnel and engineering students. Global Journal of Engineering Education, 1998. Vol. 2(No. 1).
[8]
Martin, R., et al., Engineering graduates’ perceptions of how well they were prepared for work in industry. European Journal of Engineering Education, 2005. 30(2): p. 167180.
[9]
Ridout, H., World Class Skills for World Class Industries: Employers’ perspectives on skilling in Australia, 2006, Australian Industry Group, The Allen Consulting Group.
[10]
Teichler, U., ed. Graduate Employment in Europe: The Employers' Perspective. Higher Education Dynamics. Vol. 17. 2007, Springer: Netherlands.
[11]
Nair, C.S., A. Patil, and P. Mertova, Re-engineering graduate skills – a case study. European Journal of Engineering Education, 2009. 34(2): p. 131-139.
[12]
ACER. Graduate Skills Assessment. 2011 [cited 2011 25,11,2011]; Available from: http://www.acer.edu.au/tests/gsa.
[13]
Bauer, F. and AGCAS. The UK's Official Graduate Careers Website. Product/industrial design : Your skills 2011 [cited 2011 22.12.2011]; Available from: http://www.prospects.ac.uk/options_product_industrial_design_your_skills.htm.
[14]
Berners-Lee, T., J. Hendler, and O. Lassila, The Semantic Web, in Scientific American2001. p. 29-37.
[15]
Gruber, T. A Translation Approach to Portable Ontology Specifications. 1993.
[16]
Staab, S. Wissensmanagement mit Ontologien und Metadaten. Informatik Spektrum, 2002.
[17]
Blumauer, A. and T. Pellegrini Semantic Web und semantische Technologien: Zentrale Begriffe und Unterscheidungen. Semantic Web - Wege zur vernetzten Wissensgesellschaft, 2006.
[18]
Monster. Recruiting Trends. 2011 [cited 2011 September 30th]; Available from: http://arbeitgeber.monster.de/hr/personal-tipps/markteanalysen/studien/recruitingtrends-2011-QA081687.aspx.
[19]
EU-StartUps. Semantic job search: JobiJoba lauches in Germany. 2011 [cited 2011 September 30th]; Available from: http://www.eu-startups.com/2011/03/semantic-jobsearch-jobijoba-lauches-in-germany/.
[20]
Trovix Monster Adds Next Generation Intelligent Search Technology with Acquisition of Trovix, Inc. 2008.
[21]
Jobsucher. Mein Jobagent. 2011 [cited 2011 September 30th,]; Available from: https://jobsucher.jobagent.ch/search.html.
[22]
Sabine. Blog. 2011 [cited 2011 http://sabineprojekt.wordpress.com.
[23]
Protégé. Stanford Center for Biomedical Informatics Research: The Protégé Ontology Editor and Knowledge Acquisition System. 2011; Available from: http://protege.stanford.edu/.
[24]
Richter, S., Teamplanung mit semantischen Webtechnologien, in Institut für Informatik2009, Freie Universität Berlin: Berlin.
September
30th];
Available
from: