Today there is an urgent need to understand which is the impact that technology is having on the workforce. This urgency is given by the paradigm of Industry 4.0, that is introducing rapid and epochal changes and challenges. Let’s understand how our B4DS team developed a semi-automatic procedure to estimate the impact of Industry 4.0 on job profiles and skills!

The whole society could be interested in understanding how Industry 4.0 is shaping workforce:

• companies that need for an alignment of their internal competencies;
• labor force that feels threatened by robots and Artificial Intelligence which are succeeding in many new tasks;
• governments that are trying to look to the future of sectors that characterize modern economy;
• universities that are reshaping their offer almost every year.

Despite the great effort of these stakeholders, there exists a lack of tools able to detect the impact that technology is having on specific jobs, since technological change is becoming increasingly challenging to measure, given its rapid (and often unpredictable) consequences on the worldwide economy.

In this B4DS Notebook, we talk about a scientific study done by our groups (Silvia Fareri, Gualtiero Fantoni, Filippo Chiarello, Elena Coli) and Whirlpool (Anna Binda), that are published in Computers in Industry journal[1]. The methodology we develop was tested in Whirlpool, that represents an excellent case study for two main reasons: it is a multinational company with an advanced system of skill mapping and belongs to white goods sector which is one of the most important in Italy and has already began its digitization process, especially thanks to its size and revenue.

To face this challenge, we design a novel methodology: once the job profiles were pre-processed, we tagged the texts searching for technologies 4.0 and techniques (using TECHNIMETRO®[2]). From these structured data, we then developed a measure able to quantify the readiness 4.0 of each job profile. The final step of the process was to consider the metadata of the job profiles with the aim of understanding if there are differences between 4.0-ready and non-4.0-ready at a skill level.

The input data were represented by a set of 78 job descriptions belonging to Whirlpool. The data contained a list of competencies and skills, required to perform a specific role in the organization

Each job profile was reported in a page only, all divided into two columns, as shown by figure 1. In the first column, we found a description of its job mission and its customized main activities (different for each profile); in the second column, we could find its skills divided in three groups (everyday execution, operational and functional). The different skills were 40 in total. Because of the heterogeneity of job missions and main activities, the amount of text varies in relation to the job profile considered.

Figure 1. An example of a job description and its structure.

Firstly, our methodology takes as input the job profiles and searches for all the industry 4.0 technologies contained in the text: as a result, for each job profile we have a list of technologies and techniques 4.0. Not all the 78 job profiles contain at least one technology 4.0, but only 20% is covered.

The figure 2 shows the first output of the analysis performed and the relationships among job profiles and different technologies belonging to TECHNIMETRO®. The connection is visually represented only if the job description contains one of the technologies.

Figure 2. Graphic representation of the technologies 4.0 related to each job profile.

The queues are interesting: all job profiles that perform Process Technology are linked to the technology Advanced Manufacturing. Vice versa, the mechatronics and simulation technologies, which are in the technologies queue, are all linked to a single profile: the Manufacturing R&D Manager. Another peculiarity is represented by the strong connection between profiles and the technology Predictive Analytics. Moreover, the image reveals that the analytics tool is mostly used in accounting and maintenance.

At this point, we split job profiles in two groups, assuming that: a job profile could be defined “4.0″, if it contains at least one technology 4.0 in the description of its main activities; a job profile could be defined “non-4.0″, if it does not contain technologies 4.0 in the description of its main activities.

Since each skill level is measured on a scale from 0 to 3, we computed the mean and the standard deviation for 6 group of skills: everyday execution, operational and functional skills for profiles 4.0 and the non-4.0 ones. The statistics for the specific skills are computed in a similar way with respect to the group. In fact, we computed the mean and the standard deviation for 80 group of skills: each of the 40 skills for 4.0 and the non-4.0 profiles.

We highlight the most interesting insights of this analysis:

• all the skills belonging to the Everyday Execution cluster are transversal and many of them are methodological (e.g. Project Management and Change Management). In this group, particularly relevant are Business Acumen, Project Management and Change Management, in which the gap between profiles 4.0 and non-4.0 is substantial;
• for the skills belonging to the Operational cluster, the job profiles non-4.0 have a higher score than the 4.0 ones. In particular, the Customer Quality and Continuous Improvement are averagely mastered at a higher level by job profiles non-4.0. Instead, skills such as Innovation and Price and Margin Realization remain the prerogative of professional profiles 4.0. Therefore, the job profiles 4.0 are stronger on Innovation and Price & Margin Realization; it is very consistent with their role as “managers and users” of emerging technologies;
• for the skills belonging to the Functional cluster, Job profiles 4.0 present a higher average level of skills such as Process Technologies, Equipment Design, Factory Info Systems, Ergonomics and Maintenance Expertise. For example, the average level of the skill Maintenance Expertise for profiles 4.0 is higher than its distribution peak level for profiles non-4.0. In general, all these skills require the knowledge of technologies 4.0 to be better applied. It seems consistent, because they are strictly related to the technologies: the most fitting example is the skill Factory Info Systems, or the ability to manage the hardware and software needed to collect, process, and share data inside and outside the organization.

Our classification of job profiles (4.0 and non-4.0) was then validated through the information available on O*NET (The Occupational Information Network), which is an open-source database developed for the U.S Department of Labor, made of 974 occupations from Standard Occupational Classification (SOC). O*NET categorizes a job as bright outlook if it is expected to grow rapidly in the next several years or will have many job openings.

For each Whirlpool Job Title 4.0, we searched for its correspondent occupation on O*NET, extracting the number of hot technology skills owned and tracking if it was labelled as bright outlook or not. We did the same for 8 random profiles categorized as non-4.0, to double check our outcomes.

The validation process returned 21 of 23 matches (91,3%). Generally, the profiles labelled with bright outlook are characterized by a consistent number of hot technology skills (higher than 15), that was in line with our expectations. The presence of a job profile wrongly labelled as non-4.0 indicates a technology that is currently missing on Whirlpool database. This underlines the necessity to update Whirlpool Job descriptions, that should be aligned with Competence Frameworks and the ever-changing labor market language.

The proposed tools could be used by HR managers (main beneficiaries of this work) to collect information about key professional figures 4.0 and skill 4.0 gaps of the company. This gives to the manager a data driven map of the competences 4.0 in the company and makes it evident which is the impact that technologies 4.0 are having and will have soon. Considering this map, different actions can be taken to start a data-driven improvement of the HR management process.

Let us know if you have suggestions or comments! b4ds.unipi@gmail.com

By Elena Coli and Vito Giordano

[1] Fareri, S., Fantoni, G., Chiarello, F., Coli, E., & Binda, A. (2020). Estimating Industry 4.0 impact on job profiles and skills using text mining. Computers in industry, 118, 103222.

[2] Chiarello, F., Trivelli, L., Bonaccorsi, A., & Fantoni, G. (2018). Extracting and mapping industry 4.0 technologies using wikipedia. Computers in Industry, 100, 244-257.