Essnet Big Data Specific Grant Agreement No 1 (sga-1)
Download 0.99 Mb.
|
- Navigate this page:
- 3.5 Technical structure
- 3.6 Activity of the portal in more than one country
3.4 Variables availableThe information provided regarding the characteristics of the jobs offered differs largely from one site to another. More variables provide an opportunity to produce richer statistics and so sites that give access to more variables should be preferred. As a first step, the pilot focuses on the structured (or semi-structured) information available on the job portals. The full job advertisement may provide further information, but this would need to be extracted from the unstructured text by the use of text mining procedures. Job characteristics that can be found on job portals include the following:
Further items sometimes include supervisory functions or the salary, the latter however depends a lot on national circumstance. Many job advertisements on UK portals provide salary, but this is very rare in Germany and Sweden. 3.5 Technical structureEven if the variables detailed in section 3.4 are available on a given job portal, their usefulness for statistical production depends on the way the information is provided on the web site. A crucial aspect is which variables are available in a structured format on the website. Most job portals return a summary list of search results, which only shows a limited range of variables. A list of search results usually include the job title, the employer (or a logo of the employer), the location and the date of the advertisement (see Figure ). Figure : Example for a typical list of search results 
Source: http://www.monster.co.uk/jobs Further information might be obtained using filter or advanced search functions offered by the job portals that provide the opportunity to select, for example, only full-time posts or only specific occupations. The algorithms behind these filters are however not transparent, and it is not clear whether the filtered results exhaustively represent all job advertisements posted on the job portal. The analysis of how much filter or advanced search functions can be used for web scraping is one of the major issues to be analysed in the next stage of the pilot. The entries on the list of job advertisements found are linked to job advertisements that provide more detailed information. Depending on the job portal, there is however a large variation of practices which may have strong implications on how much effort needs to be taken to use a job portal for statistics production:
Figure : Additional information specified on a standardised second level of the list of search results 
Source: http://www.stepstone.de/jobs
3.6 Activity of the portal in more than one countrySeveral of the job portals have web sites in several countries in Europe as well as in non-European countries. For example, Monster.com has job portals in more than 40 countries4 and Stepstone.com runs job portals in six European countries.5 Also job search engines like indeed.com or Adzuna.co.uk have national web sites in various countries. Since the technical structure of the web sites of the same provider in different countries often seems to be similar, one may want to make the presence of a job portal in several countries a criterion for its selection, in order to be able to reuse the procedures developed in one country also in other countries. However, since most job portals originally started in one country, their importance can vary strongly from one country to another. For this reason, the fact that the URL sounds similar does not necessarily imply an increase of international comparability to use data from the same portal also in other countries. Still, experiences gained in one country with a job portal should equally be used in others interested in using data from this portal. Download 0.99 Mb. Share with your friends:
|