This report has two focus areas. The first area described is the knowledge base of cases created to populate the Intelligent Decision Support System (IDSS) developed with the ENLIVEN project to support evidence-based policy making within participating EU countries, with an emphasis on young adults who are Not in Education, Employment and Training (NEET). The second area considers a detailed study of an employment support programme for young people (the Young and Successful project), conducted to gain a deeper understanding of the barriers young people encounter on their journey to employment.
Within the first area a glossary of relevant terminology to the IDSS is presented. The IDSS is built in ENLIVEN using the case-based reasoning (CBR) methodology. CBR is conducted based on two key mechanisms which discover and model the domain knowledge:
- Case representation, which is a unified format (i.e. attributes) to store cases in the case base. An explanation is given of the methods used to derive the case representation. A unified template of 78 attributes is established to represent and store details of NEET projects as cases. Documentation sources for the NEET cases are described. A case base of 77 cases has been prepared (30th September 2018) using the consistent case representation with 78 attributes.
- Similarity measure based on case representation, which calculates how similar two cases are. In CBR systems, not all attributes can contribute to retrieving the most relevant or useful cases in terms of supporting informed decision making. An initial set of four key attributes, namely Location, Aims, Target groups and Activities are identified as the key attributes to determine similarity. Methods used to build the similarity measures for the four key attributes, namely clustering analysis and knowledge acquisition from experts are described. An explanation of similarity measure algorithm is given.
Based on the case representation and similarity measure, a description of the operation of the initial knowledge base is presented. Transferability of the IDSS to other policy areas is discussed.
To study the Young and Successful project (YaS) two data mining approaches were utilised: proximity to the labour market indicator analysis and decision tree analysis (a machine learning technique in Artificial Intelligence) of programme participant employment outcomes. An important question considered was whether the decision tree analysis would support anecdotal experience from local practitioners which indicated the relevance of Maslow’s hierarchy of developmental needs to their work within the context of young people furthest from the labour market. It was found by decision tree analysis that young people who possess good levels of confidence, mental health and who have worked before are more likely to secure an employment outcome. The collaboration with YaS project was used to inform choice of attributes for the IDSS case base. The YaS project will be stored as one of the cases in the knowledge base.
The report concludes with two recommendations:
- A consistent standard of documentation within a coherent framework across countries for future development in research and practice is required.
- More research schemes supporting inter-disciplinary research within a standardised infrastructure are needed.