Mapping career causeways with the Sussex Chamber of Commerce

We have built a demonstrator tool that is linked to the Future Skills Sussex website that leverages the algorithm described in our Mapping Career Causeways report to help career advisors in Sussex recommend suitable career paths to their advisees.

Introduction

The Sussex Chamber of Commerce (CoC) has launched a programme called Future Skills Sussex aimed at improving the local economy with homegrown talent. As part of this effort, the Sussex Chamber sought to develop a tool for career advisors to aid job seekers with career transitions, especially into sectors identified as high priority for the local economy.

In November 2020, Nesta published a report titled Mapping Career Causeways: Supporting Workers at Risk describing a system we developed for supporting job transitions in a changing labour market. As part of that system, we built an algorithm that, for a given job, can identify ‘similar jobs’ based on a range of criteria (specified below). The motivation for creating the algorithm was the distinct lack of any official data on the movements that workers in the UK make between occupations. In the absence of this data, the algorithm captures the theoretical similarity between jobs and provides a sense of the feasible transitions between roles.

We have recently launched a partnership with the Sussex CoC to build a demonstrator tool to illustrate how the algorithm we developed for Mapping Career Causeways could be adapted to support the Future Skills Sussex programme.

Algorithm overview

The Mapping Career Causeways algorithm measures similarity between occupations. It was built from two open frameworks that describe the features (such as skills) that are required for a range of occupations. The European multilingual classification of Skills, Competences, Qualifications and Occupations (ESCO) is a hierarchical ontology that maps over 13,000 skills to over 3,000 occupations relevant to the EU labour market. We supplement the ESCO occupational data with additional information from the O*NET database, developed by the United States Department of Labor.

Similarity is evaluated based on four features of the occupations:

1. essential skills: skills that are required for each of the occupations (eg, project management, process data, analyse economic trends)
2. optional skills: skills that are desired for each of the occupations (from the same list list as essential skills)
3. work activities: the common activities required for the occupations (eg, interacting with others, supervising people) (corresponds to level 2 of the ESCO skills pillar)
4. work contexts: factors that influence the nature of the work — these can be physical (eg, frequency of exposure to job hazards), structural (eg, duration of typical work week), or describe interpersonal relationships (eg, coordinate or lead others)

The diagram in Figure 1 shows how these features are combined to calculate similarity between occupations.

Figure 1: Algorithm architecture

The tool

We have built a demonstrator tool (SkillsMatcher) that is linked to the Future Skills Sussex website that leverages the algorithm described above to help career advisors in Sussex recommend suitable career paths to their advisees.

Figure 2 shows what the user sees when first entering the tool. The tool prompts the user to input a job (pulling from the list of approximately 3,000 occupations defined by ESCO) to use as a starting point. The user can choose to search specifically within one of Sussex’s high priority sectors (Manufacturing and Engineering, Digital, Health and Care, Construction, Visitor and Hospitality, Creative and Cultural, and Land Based) or to show all viable transitions. The user can also select how many matches to show (between 1–15). Based on these inputs, the tool filters a pre-calculated similarity matrix between all occupations (created using the algorithm described in the previous section), to display the most similar occupations. The more results that are shown, the poorer quality they are, as the most similar results are presented first.

Figure 2: Landing page for our demonstrator tool

The tool also provides additional information along with each match (shown in Figure 3), to explain to the user how to make the transition. For example, we note all of the essential skills that we think the person already possesses based on the starting job, and all of the skills we think they would need to acquire to transition successfully to the new job.

Figure 3: Detailed information accompanying each recommendation

Strengths and Limitations

Strengths

The algorithm that we developed for the Mapping Career Causeways project is not geographically specific, yet it was able to be adapted to meet the needs of the Sussex CoC with minimal changes. The backbone of the algorithm (the similarity calculation) is extremely generalizable, and additional customisations (such as filtering by sector) are easy to add. It is also highly scalable, despite being developed for a standalone report, it is able to be leveraged to make predictions on the fly within a reasonable time for a user facing app.

Limitations

There is currently no UK framework of occupations that describes the skills and other features of occupations. In the absence of such a framework, we had to use frameworks from the EU and the US as the basis of our similarities. As a result, not all of the jobs are relevant to the UK (ex: Senator vs Member of Parliament). We also relied on manually generated crosswalks between these frameworks which are prone to error and do not necessarily capture all equivalencies.

The tool is also not necessarily meant for making job transition recommendations, it exclusively measures similarity between jobs. There are many reasons why a “similar” job may not be one that would be considered a viable transition (eg, the level of education/experience required, the salary, or the location).

Opportunities for Nesta’s missions

Although this particular demonstrator tool was developed in partnership with Sussex for a specific need, it can easily be adapted in support of Nesta’s missions.

For example, the fairer start mission is interested in strengthening the early years workforce. We could potentially repurpose the algorithm to build a tool focused on identifying similar jobs in the early years sector. Likewise, the sustainable future mission is interested in defining and promoting green jobs. The tool could be expanded to search specifically for transitions into green jobs and sectors. We could also highlight the green skills and work activities that are required in any role.