Outils personnels
Vous êtes ici : Accueil Publications Learning Technologies Proceedings CV Interoperability Seminar 2007 CV 2007 Outcomes

CV 2007 Outcomes

On 18-19 June 2007, in partnership with HR-XML and Liberty Alliance, EIfEL organised an international seminar on CV interoperability in Paris. Representatives from European agencies, national public employment agencies, job boards, governments, international and European standardisation bodies, service providers and academics contributed to a series of rich exchanges to reflect on the need for, the benefits of, and the solutions to CV interoperability: how to construct, disseminate, match and exploit curriculum vitæ information to support mobility, employment and innovation?

As the world of CVs is characterised by a series of different specifications, one of EIfEL's aims is to contribute to their convergence (e.g. HR-XML and Europass CV) into one international standard. Meanwhile, it should at least be possible to transform a CV written on one system using one specification in order to be read by another system using a different set of specifications. This is precisely what the new web service developed by EIfEL does. It was revealed by Marc Van Coillie, EIfEL's CTO during the seminar: a CV 'translator' capable of transform a CV based on one specification into another.

One of the other main outcomes of this seminar is the creation of a working group whose initial task is the creation of a HR-XML data exchange service using Liberty Alliance Federation of Identities and Federation of Services to support secure exchange of human resources information while respecting privacy. This work will be then extended to other digital identity management services, such as OpenID and CardSpace.

The curriculum vitæ challenges

While job applicants, employers, job boards and public employment services exchange millions of curriculum vitæ each year, very few of the systems managing CVs are yet able to share information in a meaningful way. Increasing the level of interoperability between the different stakeholders could lead to far better service and huge savings. With agreed structures and vocabularies, the semantic value of the information contained in CVs will be greatly increased.

This is the work undertaken by HR-XML: providing human resource professionals with the basic specifications to support a variety of HR processes, from recruitment to managing sub-contractors. But are HR-XML specifications enough to guarantee full interoperability across HR systems? Are two systems HR-XML-compliant fully interoperable? The reality is that these specifications only provide a framework from which can be derived many different implementations, not necessarily compatible one with another. In order to be fully interoperable, systems need to go one step further in defining application profiles i.e. an agreed set of constraints that will reduce the number of "interpretations" of the basic specifications.

Some of the other key challenges are the issues of privacy (how can a person control who can see what of his/her CV – e.g. anonymisation?), currency (how to keep one's CV updated?) and authenticity (how can we be sure of the claims made?). The contribution of Liberty Alliance to the seminar was critical for exploring how federation of identities and federation of services (under end-user control) could provide a framework to find possible solutions to these challenges.

State of the art review

In order to make documents easily readable by a machine, either for interpretation or simply to change the layout for publishing purposes it is necessary to enrich the semantic information of the data contained in a CV. We need to mark up the text to flag fields such as "address", "work experience", "skills", etc.

What the seminar made clear is that the semantic representation of CV data is taking two different, and possibly complementary, directions:

  • XML schemas (externally referenced to documents) – this is the way chosen by HR-XML, Europass and a number of providers. The main advantage of this approach is that as all documents provide a reference to the base (or derived) schema, the interpretation of data is clearly defined. It is  also relatively easy to test whether or not a document is conformant to the specification, hence better interoperability prediction. Implementation requires professional programming skills.
  • Microformats (internal to the HTML code) – this is the way chosen by some social software providers. The main advantage of this approach is its accessibility to the non-expert: microformats are a simple extension of HTML (XHTML) and can be implemented using simple style sheets (CSS) where special classes of new HTML objects are defined (e.g. "skill" or "knowledge", etc.). The lack of external reference makes it more difficult to ensure consistency of documents against the specification.

The first one could be designed as 'top-down' while the other would be 'bottom-up' or 'grassroots.' One is mainly supported by industry that has set-up standardisation committees, while the other one is more the result of the consensus among actors who want to find quick solutions to the same problems with a focus on easy implementation.

A CV is also a document that contains personal data, and its author should be able to control who has access to what, how and when. This type of control requires identity management services (IDM), which comprise identity federation (the ability to federate several accounts in order to provide single sign on – SSO – i.e. limit the number of passwords to remember) and federation of services (the ability to share profile data across different accounts – typically having one's address at only one place so any change just have to be propagated to the federated services).

It is interesting to note that, as in the field of CV semantics, the field of digital identity management displays two different, and possibly complementary, approaches:

  • Liberty Alliance – a consortium that has developed a set of standards, widely adopted by telco operators as well as states like Norway or France to support eGovernment initiatives. LA provides a framework for IDM (identity management) that includes federation of identity and federation of services across organisations.
  • OpenID – a consortium of individuals and organisations that has designed a technology for allowing individual to federate identities and share profile data across services.

Both frameworks offer more or less the same types of services – the degree of advancement and completeness of these services varies today and will probably converge in the near future. The main difference is that one (LA) requires an agreement between organisations who wish to belong to a "circle of trust" while OpenID can be implemented without any policy agreement across organisations. In order to allow someone to use an OpenID, a service provide only has to add a few lines of codes and does not have to enter into policy negotiations with other organisation..

In terms of complexity and implementation, both solutions are relatively easy to implement. The Lasso library, developed by Entr'ouvert, a member of EIfEL, can be easily implemented to include existing information systems into the LA framework. The IDcommons portal provides several solution for OpenID.

LA is a very successful "top down" initiative, that has focused a narrow problem with a big spectrum of applications (identity management) and has produced widely adopted specifications and protocols. OpenID is a very successful "bottom-up" initiative that has even been adopted by large organisations such as America Online.

From these two examples (CV semantics and ID management) one should recognise that there is not just one way to do things and that bottom-up and top-down approaches have their strenghts and weanesses and could even be complementary.

As part of the Concordia initiative of Liberty Alliance, some of us are currently reflecting on a framework where LA and OpenID would coexist: LA at the organisational level, OpenID at the individual level. Someone could thus log into a system using his/her OpenID, and once logged in could access a federation of services through a LA identity. Concurrently EIfEL is exploring how microformats could provide a "standard" solution for rapid prototyping of specifications that could then be abstracted into XML schemas, once stabilised, to provide a reliable foundation for interoperability and conformance.

Current issues

A number of issues and use cases were presented during the seminar:
  • Call for tenders (presented by Unisys): the response to a call for tenders generally includes the CVs of key. For the candidate: CVs can be out of date and penalise the outcomes of the review process; time can be lost in rewriting them or putting them in the right format. For the client, CVs can be difficult to exploit manually if they do not share the same structure or layout; it would be useful if CVs of candidates for tenders could be exploited like the CVs of individual candidates.
  • Multiple CVs publication: the same person can apply for a series of jobs, each one with its own requirements, hence the need for different CVs to elicit the required competencies. Each application may be done on different systems, with their own idiosyncrasies. Candidates could save precious time if it were possible to repurpose the same information, and publish it in a format that could be imported to many different employment sites.
  • Privacy and anomysation: a person may wish to hide his/her true identity either because of current position/employer and/or because of such social aberrations as ageism, racism or misogyny. This include the anonymisation of the CV itself as well as the attachments and references such as certificates, diploma supplements etc.
  • Job matching: how can we match one or a pool of CVs with one or a pool of job offers?

Current developments


CEDEFOP, the European agency for vocational education, is providing a series of services to support CV interoperability for improved mobility of the workforce. It is based on the vision that everybody should be responsible for editing and distributing their CV, the role of CEDEFOP being to provide technical specifications and services for editing and publishing a CV readable by people and machines.

The most developed services currently comprise:
  • An online editor of CV – that can be saved on one's computer and re-imported in the editor for further editing
  • A Webservice to dynamically create documents (PDF, Word and OpenOffice) from XML data – a service exploited by EIfEL in its 'translator' and EURES, the European public employment agency toexport documents
  • A CV in a PDF format with the XML attachment, so the file can be read by a human and a machine altogether.

Today, nearly 400,000 people hae visited the Europass site. 6,000,000 documents were downloaded and 700,000+ CVs were created in 2006 (300,000+ in 2005).

Europass-CV is currently planning to evolve its current specification towards HR-XML which provides greater semantic value – several HR-XML named fields are simply generic text fields in Europass.

During the seminar Kite and EuroCV.eu were also presented as platforms supporting the publication of Europass conformant CVs. Kite, a CV editor plug-in for blog platforms was presented as an alternative for editing and distributing CVs, using blog services to manage access rights. The CV created can be exported both in Europass and HR-XML Europortfolio formats.

HR-XML Europortfolio.

EIfEL, as leader of the HR-XML workgroup on CV interoperabilityn has produced a résumé specification, HR-XML Europortfolio that aims is to bridge the gap between the previous HR-XML Résumé specification, Europass and IMS ePortfolio. The first version of the Europortfolio specification focuses on the résumé dimension of an employability ePortfolio and will cover the other fields in future releases.

The seminar has confirmed an increased adoption of HR-XML specification for résumés handling, for example GermanCV, iProfile UK (2 million CVs). Each implementation tends to develop its own application profile (a restriction and/or extension of the base specification) which might create interoperability problems, but as EIfEL's 'Translator' has demonstrated, it was possible to transform one format into another with minimum loss of semantic information. Further analysis has shown that both profiles share over 80% of information with the Europortfolio specification and that harmonisations should be possible.

It is interesting to note that some of the differences are linked to cultural differences in the handling of applications. For example the GermanCV is not interested in the detailed information Europortfolio specifications (based on Europass) provide on languages competencies. German recruiters are only interested in summary information on linguistic level as the selected candidates will have to face a formal assessment of their linguistic abilities. The detailed linguistic information of Europass seemed more meaningful for developing a reflexive approach to one's linguistic competencies than for recruitment.

The future HR-XML 3.0 specification will completely revise its structure. In order to avoid redundancy of elements across different specifications (like candidate and résumé), the structure is now decomposed into a series of basic components (kinds of speclets or micro-specs) that can be aggregated as needed to construct a specific profile. Meanwhile the latest release (2.5) describes diplomas according to the ISCED classification.

hResume microformat

Microformats are an alternative to XML to enrich with semantics the information contained in a document by tagging its different components to make it readable both by the human (display format) and the machine (semantics coding). hResume is the microformat specification developed specifically for enriching the data contained in a résumé.

Here is an example of a skill microformat (defined in a standard HTML 'class') which is simply a link:
I have skills in <a class="skill" rel="tag" href="http://en.wikipedia.org/wiki/Bow_%28weapon%29">bow hunting</a>
and <a class="skill" rel="tag" href="http://en.wikipedia.org/wiki/Nunchucks">nunchucks</a>.
During the seminar we learned that some of those who use the hResume microformat have developed their own vocabulary, thus reducing interoperability.
While microformats are still at very early age for, the approach seems worthwhile exploring as complementary to XML.

Towards a European repository of competencies?

One of the recurring issues when describing one's skills and competencies is the ability to find a taxonomy that is detailed enough so that, instead of writing a string of characters in the description of the competency, it would be possible to use a URI (unique resource identifier) that would point to an agreed description of the competency. In doing so, it would be possible to add a multilingual description of the competencies which would facilitate mobility of the workforce.

The DISCO project, with its detailed and multilingual taxonomy of competencies, has demonstrated a possible solution to this problem. Seminar participants discussed the possibility, in the near future, to develop a web-service that would provide an XML stream of competency description in response to a query of a CV editor.

Digital identity, social networks

Today there are people who belong to different social networks, have different spaces where they manage their identity, not to mention the different places where CVs have been edited and sent. How is it possible to manage all those different identities?

ZIKI presented its solution for managing and aggregating one's various identities, including the publication of a CV based on data contained in the LinkedIn public profile – which uses hResume microformat. This aggregation of information will certainly facilitate the maintenance of heterogeneous and distributed personal data. It is possible to imagine that a CV will be built out of the aggregation of information contained in different repositories.

Securisation of attribute exchange across services

Enforcing privacy is one of the key issues in the management of CVs – e.g. anonymisation. If now we want to move towards an architecture where people have full control, from a central location of their personal data and CVs, one of the issues to address is the securisation of attributes exchange across services – e.g. personal data being sent to a job board, then an employer, then to a firm for background checking.

The Chief Architect of Symlabs proposed to develop a new Liberty Alliance web service for human resources – an HR ID-SIS in LA's terms. This new service would use combine the LA privacy services with the HR-XML data model to provide a secure transport layer for personal and HR data. A working group has been created to develop the first demonstrations for the next Plugfest, during the HCSIT-ePortfolio Summit, next October.

Towards a standardisation of a European CV?

Led by the CEN (European Center for Standardisation) pre-standard committee Director, one session explored the benefits and possibilities of moving from the emerging de facto to de jure standards, through a formal standardisation process of résumés in Europe.

While no concrete steps have yet been agreed, the proposed paths were:
  1. A Common Workshop Agreement (CWA) around the Europass profile as part of the current workshop on learning technologies. This could lead to the publication of a series of recommendations and guidelines that European countries would be free to implement – or not.
  2. The submission of this CWA to a Technical Committee, in order to make it a formal standard that would be binding for all the member countries. National members of CEN (e.g. French AFNOR or German DIN)would publish and enforce the standard.

Provisional conclusions

It appears that we currently have three different, but possibly complementary, approaches to supporting representations of the résumé that are understandable both by human and machine:

  • (X)HTML microformats, where specific tags (for the machine) are added to HTML (for the human) code – this is the solution proposed by some social networks and Web 2.0 tools.
  • XML (for the machine) that can be transformed into HTML (for the human) with XSL (transformation sheets) – this is the solution proposed by HR-XML
  • PDF (for the human) with an XML document attached (for the machine) – this is the original solution proposed by CEDEFOP

EIfEL was able to demonstrate with its Translator web service using a series of XSL transformations that it is possible to move from one format to another, including the different profiles such as GermanCV, iProfile and the French Universal CV (CV Universel). The translation uses Europortfolio, the HR-XML representation of a CV, as the 'standard' representation of a CV.

We believe that in the near future Europass-CV and HR-XML Europortfolio will converge. This will increase significantly the semantic value of Europass with the ability to represent competencies and relations between CV parts and annexes – this could be evidence supporting the claims made in the CV. HR-XML Europortfolio will in turn include the elements required to support the European language portfolio of CEDEFOP. We should also study conjointly the integration of optional elements for the CV of researchers.

But the value of the information contained in the CV will only be fully exploited if we are able to define protocols to transport personal information safely and we can ensure simple mechanisms to keep the information current. This will be one of our main tasks for the next few months, to be ready for the Plugfest and the Human Capital and Social Innovation Technology Summit, 16-19 October.

For more information, please contact Marc Van Coillie: marc.van.coillie@eife-l.org


Actions sur le document