 |
e-recruitment, competency model, ontology, semantic web, annotation, web-based knowledge management
The web has drastically changed the online availability of data and the amount of electronically exchanged information. It has revolutionized access to personal information and knowledge management in organizations. E-recruitment is one of the typical application of such a knowledge management approach through the web. The traditional recruitment methods of advertising vacancies in newspapers, magazines and trade press, as well as advertising at job fairs, employing recruitment agencies and registering with search firms were adequate in the past. Access to the Internet has proven that these methods are too slow, expensive and lacking in their ability to deliver high-quality candidates in the shortest possible time in the modern employment market. The e-recruitment market is currently booming; it will reach 13.4 billion of dollars by 2005, with a compounded growth rate of 53 percent.
Faced with this explosion of e-recruitment, many candidates deposit, each day, their Curriculum Vitae (CV) in what is commonly called a CVBank (the greatest websites announce until more than 50.000 CVs recorded in their bases). In addition to a broader diffusion, this approach offers to a job seeker both reactivity and facility when updating its personal information. On their side, in addition of having a plentiful `` fishpond of candidates '', the recruiters benefit the same aspects: massive diffusion of their job offers, rapidity when contacting candidates, reactivity, etc. Moreover, for certain companies, publishing on line their job offers is a sign of good economic health; e-recruitment thus becomes a vector of institutional publicity.
Paradoxically, this revolution of the employment market based on the use of information technologies is not accompanied by an evolution of the tools dedicated to the retrieving and the management of CVs and job offers. Indeed, the techniques used to analyse the information resources related to e-recruitment remain very rudimentary. For instance, the search engine provided by Monster (http://www.monster.com) is simply based on combinations of locations, list of keywords and sectors (for instance, Human Resources or Information Technology). This approach, which seems too poor for being really effective from a qualitative viewpoint, is prejudicial both for recruiters and for job seekers. The former have difficulties to choose the best person for their needs (`` recruiters are under a rain of CVs and information whereas they are thirsty for competencies ''). The latter have difficulties to emphasize their competencies through the writing of their CVs (and thus to break free from competition) and have difficulties to find job offers, which ideally correspond to their profiles.
The goal of the CommOnCV project is to provide new job matching services based on competency management. The principle underlying CommOnCV consists in considering a CV (respectively a job offer) as a synthetic view (expressed, in natural language, in terms of qualifications, work experiences and extracurricular activities) of a richer network of competencies. According to this principle, the first objective of the project is to allow the end-user (i.e. a job seeker or a recruiter) to make all the competencies underlying its resources (i.e. its CV or its job offer) explicit. The second objective is to formally represent these competencies in order to provide more powerful e-recruitment services: the content (expressed in terms of competencies) of CVs and job offers must be manageable by computers in order to provide automatic matching processes.
These objectives require (1) the definition of a competency model and (2) the definition of a process dedicated to the management (i.e. identification, formal representation and exploitation) of the competencies underlying a CV or a job offer.
The competency model we advocate is based on the following definition: `` a competency is the effect of combining and implementing resources in a specific context (including physical, social, organizational, cultural and/or economical aspects) for reaching an objective (or fulfilling a mission)''. Three types of resources are distinguished: (1) knowledge which includes theoretical knowledge (e.g. to know the second law of thermodynamics), knowledge on existing things (e.g. to know the operation of test-bed equipment for combustion engines), and procedural knowledge (e.g. to know the mounting procedure of a particular electronic card), (2) behavior which includes human trait, quality and attitude (e.g. leadership, integrity, persuasiveness or adaptability) and (3) know-how which includes formalized know-how (e.g. the application of working procedures) and empirical know-how (e.g. trick, ability or talent). A detailed description of this model is proposed [1].
The process we advocate is based on techniques currently developed in the semantic web area. These techniques are concerned with (1) the building and sharing of domain ontologies and (2) the annotation of electronic documents.
In our work, domain ontologies are used as reference systems for identifying the competencies respectively required for a job and acquired by a person. They provide all the concepts (and their relationships) that are necessary for representing the resources, the context and the objective of a competency. These concepts can be related to tasks associated to a job position, know and know-how underlying a diploma or a task, organizational cultural or economical aspects of a firm, etc. Several types of ontologies can be differentiated, in particular sector ontologies (e.g. Finance/Banking or Healthcare) which provide a fine and inflexible description (reached by consensus) of the job positions (and their associated tasks, knowledge, skills and abilities) of the considered sector and entreprise ontologies which include more singular aspects of firms such as the specific tasks exclusively performed by an enterprise or the technological and economical description of this enterprise. We are currently working on an ontology dedicated to the sector ``Information and Telecommunication Technologies'' (ITT). This ontology is defined from an ITT job position reference system provided the Cigref (http://www.cigref.fr), an ITT Club of the major French corporations such as Air France, France Telecom or Paribas. In the context of CommOnCV, ontologies are crucial because they allow a recruiter and a job seeker (i) to have a common understanding of the competencies and the tasks underlying a profession and (ii) to share the same vocabulary for denoting theses notions.
Note that given a sector, several organizations can assert their own reference system and these reference systems can evolve quickly (evolution of job positions is sometimes an intrinsic characteristic of a sector). In this context, it will be necessary to study new topics of interest (related to ontology engineering) such as ontology evolution and coordination.
In our work, the electronic documents that are annotated are CVs and job offers available in the current websites dedicated to e-recruitment (e.g. Monster or Jobpilot). The intended annotations aim at making all the competencies (resp. acquired and required) underlying these documents explicit. The annotating process we advocate is based on the use of ontologies constructed from job position reference systems. Figure 1 illustrates the result of this process in a context of a work experience described in a CV. For a CV (resp. job offer) in its entirety, this process produces a set of annotations characterizing all the competencies acquired by a person during its history (resp. required for a job offer). These competencies are complementary from the linguistic expressions used to illustrate them in a CV (resp. job offer). Formally represented by using languages such as RDF/RDFs or OWL, these annotations are used (thanks to the inference mechanisms provided by the languages) to facilitate the matching process between ``supply and demand''.
The ``competency logic'' is not a new approach in Human Resources Management; this logic is now practiced in most of organizations for different purposes such as staff development and deployment, job analysis, learning organization or economic evaluation. However, in the context of e-recruitment, this logic is not yet applied. The advent of the semantic web must be considered as an opportunity to develop this logic for e-recruitment. CommOnCV must be considered as a first attempt to this purpose.
In order to validate our approach, several prototypes such as CommOnReference or CommOnAptitude are currently developed. CommOnReference (developed with Protege-2000) is a knowledge acquisition tool dedicated to the building of ontologies defined according to Employment Reference Systems. CommOnAptitude is dedicated to the identification of behavioral aptitudes (such as Ability to communicate orally or Ability to manage conflicts) from personal traits (such as integrity or persuasiveness); this tool is defined in collaboration with the psychologists of a French firm which develops softwares dedicated to human resources analysis.