Multi-generational labour markets: Data-driven discovery of multi-perspective system parameters using machine learning

Abstract

The impact of aggressive capitalist approaches on social, economic and planet sustainability is significant. Economic issues such as inflation, energy costs, taxes and interest rates persist and are further exacerbated by global events such as wars, pandemics and environmental disasters. A sustained history of financial crises exposes weaknesses in modern economies. The Great Attrition, with many quitting jobs, adds to concerns. The diversity of the workforce poses new challenges. Transformative approaches are essential to safeguard societies, economies and the planet. In this work, we use big data and machine learning methods to discover multi-perspective parameters for multi-generational labour markets. The parameters for the academic perspective are discovered using 35,000 article abstracts from the Web of Science for the period 1958–2022 and for the professionals’ perspective using 57,000 LinkedIn posts from 2022. We discover a total of 28 parameters and categorized them into five macro-parameters, Learning & Skills, Employment Sectors, Consumer Industries, Learning & Employment Issues and Generations-specific Issues. A complete machine learning software tool is developed for data-driven parameter discovery. A variety of quantitative and visualization methods are applied and multiple taxonomies are extracted to explore multi-generational labour markets. A knowledge structure and literature review of multi-generational labour markets using over 100 research articles is provided. It is expected that this work will enhance the theory and practice of artificial intelligence-based methods for knowledge discovery and system parameter discovery to develop autonomous capabilities and systems and promote novel approaches to labour economics and markets, leading to the development of sustainable societies and economies.

Introduction

Economic issues, including inflation, energy costs, taxes and interest rates, persist in our daily lives. These challenges have been amplified by global events such as the COVID-19 pandemic, environmental crises, geopolitical conflicts and wars. ¹ Global inflation, climate change's impact on economies, rising labour expenses, European gas supplies and food security are major concerns. ² Additionally, the ‘Big Quit’ or ‘Great Resignation’, characterized by a substantial number of people leaving their jobs, further compounds these economic challenges.^3–6

A history marked by financial crises, like those in the 1970s and 2008, and more recently, the FTX collapse, has exposed substantial vulnerabilities in modern economies.^7,8 The financial system, regulatory structures and economic policies are all affected. ⁹ Furthermore, aggressive capitalist approaches have strained social and environmental sustainability. To safeguard our societies, economies and the planet, transformative economic strategies are imperative.^10,11 Labour markets, integral to the economy, significantly impact individuals, businesses and society.^12–14 Investigating labour markets informs us about how these elements interconnect and how different policies and conditions can influence employment, wages and overall economic outcomes.^15–17 Research in this field aids our comprehension of evolving labour dynamics, employer roles and societal influence in labour markets.^18–20 Moreover, it allows the alignment with national and global priorities, such as the UN Sustainable Development Goals, facilitating the development of sustainable cities and societies.^21–23

Labour markets are inherently multigenerational, encompassing individuals from various age groups^24,25 (see Table 1). Generations (first, second and third generations) could also refer to generations of immigrants in the US or other countries. This diversity presents a challenge for many organizations as the generational experiences of employees influence their work tendencies. ²⁶ Factors such as technology, economic status, gender, education and ethnicity further shape work attitudes.^27,28 Recognizing these generational and contextual disparities is crucial for organizations to maintain a productive and engaged workforce. ²⁹

Table 1.

Different cohorts of generations.

Cohort	Year	Age in years (2022)
Alpha Generation	(2010–2024)	Less than 12
Generation Z	(1995–2009)	27 -13
Millennials/Generation Y	(1980–1994)	42–25
Generation X	(1965–1979)	57–43
Baby Boomers	(1946–1964)	76–56
Silent Generation	(1925–1945)	97–77
Greatest Generation	(1901–1924)	121–98

The study of generational dynamics in labour markets offers insights into the unique experiences and challenges faced by different age groups. ³⁰ This knowledge facilitates informed decision-making to support the diverse needs of workers, including younger employees dealing with issues, e.g. student debt and competition, and older workers facing challenges such as ageism and technology adaptation. ²⁶ Overall, understanding these dynamics fosters a more inclusive and supportive work environment. ²⁹

In this study, we employ big data, machine learning and advanced techniques to uncover multi-perspective parameters for multi-generational labour markets. Two distinct perspectives, one from academic literature and the other from industry and professionals on LinkedIn, provide complementary insights. Our academic dataset, sourced from the Web of Science (1958–2022), yields 15 parameters using the Latent Dirichlet Allocation (LDA) algorithm, organized into five macro-parameters. The LinkedIn dataset, comprising 57 K posts, also yields 13 parameters through LDA. ³¹ We develop a comprehensive software tool for parameter discovery, incorporating quantitative analyses and visualization methods to depict datasets, clusters and parameters. The paper presents a multi-perspective labour market taxonomy (Figure 1) supported by qualitative analysis of over 100 research articles.

Figure 1.

A multi-perspective taxonomy of multi-generational labour markets.

Previous research in the domain of this paper has employed NLP-based topic modelling for trend analysis,^32–35 addressing challenges^36–38 and sentiment analysis.^39,40 Research in social media and LinkedIn mining has focused on revealing demographic insights, ⁴¹ employee skills^42,43 and job ⁴⁴ and profile classifications.^45–51 Additionally, Twitter data has been harnessed for appraising transit service quality, ⁵² detecting traffic-related events, ⁵³ emergency management ⁵⁴ and conducting temporal trend analysis. ⁵⁵ See also.^56–59

This paper's significance lies in its contributions to understanding age dynamics, offering a data-driven multi-perspective approach and promoting artificial intelligence (AI)-driven knowledge discovery. Such insights are valuable for informed decision-making and could influence future labour economics research and autonomous systems development. A more detailed version of this article is available in an open-access archive. ⁶⁰

Methodology & software tool design

This section explains the methodology and design of the proposed architecture for the system. In Figure 2, we illustrate the architecture of our software, which is composed of four components, each of which is analyzed in more detail below. Data collection, data sources (Web of Science and LinkedIn), preprocessing, parameter modelling, parameter discovery and quantitative analysis, validation and visualization are explained in sections 2.1 to 2.5.

Figure 2.

The system architecture.

First, we introduce Algorithm 1, the master algorithm of our system. The algorithm has been called two times since we have two datasets. For the first dataset, we download the CSV file directly from Web of Science, which includes Article Titles, Abstracts, Keywords and publish date. On the other hand, for the second dataset, we collected LinkedIn posts using web scraping techniques (e.g. Python, Beautiful Soup, Requests and Pandas), as shown in Algorithm 2. The post objects were collected as JSON objects (JavaScript Object Notation). Afterward, we converted the JSON file to a CSV file.

Algorithm 1 takes the CSV file as input. Then, the CSV file is preprocessed in the following step using Pandas. Next, we used Gensim's Python library for parameter discovery using LDA. ⁶¹ Our quantitative analysis was implemented using domain knowledge, the clusters were classified as parameters, and then the parameters were grouped into macro-parameters. Our final step was to visualize the parameters and macro-parameters. In addition, these parameters were validated internally and externally.

Algorithm 1 Master Algorithm

Input: CSV file

Output: Visualization of the discovered parameters

1: CSV_file

2: Data_dataframe ← Read CSV_file

3: Processed_data ← Preprocessing data_dataframe

4: Coherence_values ← Compute coherence for different LDA models

5: Optimal_num_topics ← Find optimal number based on coherence_values

6: Optimal_model ← Load optimal LDA model

7: LDA_visualization ← Apply LDAvis to optimal_model

Algorithm 2 Data Collection (LinkedIn)

Input: search_query

Output: CSV_file

1: Link ← Load linkedin weblink

2: Driver ← Load chromedriver

3: Login_field ← Login to linkedin

4: Search_query ← Search_query

5: Driver ← Load posts page for search_query

6: Last_height ← Get the recent height of the posts page

7: while height of page keeps changing

8: New_height ← get the new height of the page

9: if New_height is equals to Last_height then

10: begin

11: exception

12: break the loop

13: end

14: end if

15: Last_height ← New_height

16: end while

17: Src ← Load posts page source code

18: Post_details ← extract description of all posts from the Src

19: Post_dict ← save post content in a dictionary from Post_details

20: JSON ← dump Post_dict

21: JSON_file ← read JSON

22: CSV_file ← create CSV from JSON_file

Data collection

We conducted an extensive literature review to identify alternative synonyms for ‘multi-generational workforce’ and used them to collect data. We used the Web of Science database to obtain the most relevant documents with an integrated query language and data format. Moreover, it promotes access to topic indexes, citation indexes and other databases from other disciplines, which can be used to find relevant research and evaluate its conclusions. Thus, we have collected our dataset from the Web of science. The dataset was generated using a Boolean query (Q1), as shown in Table 2. We have collected around 35 K articles (1958–2022) from several Web of Science disciplines, for example, Educational Research, Business Economics, Computer Science, Social Sciences, Demography, Communication and Others. The number of articles after removing duplication is around 31 K.

Table 2.

Data Collection.

Source	Querynumber	Query	Datasize
Web of Science	Q1	‘(First generation OR second generation OR Y-ers OR X-ers OR Millennial* OR Gen Y OR Generation Y OR Generation X OR Gen X OR Gen Z OR Generation Z OR baby boomer* OR multi-generation OR generation* OR Inter-generational OR cross-generational OR cohort OR demographic*) AND (workforce OR labour force OR workplace OR skill* OR Qualification* OR experience* OR work OR employment OR occupation)’	35K
LinkedIn	Q2	‘Millennials, Gen Y, baby boomers, Generation Y, Generation X, Gen X, Gen Z, Generation Z, Alpha generation, multi-generation, Inter-generational, cross-generational, Employment, Unemployment, Middle Aged, Aged, Elderly, Young, Senior, Seniors, 20 s, 30 s, 40 s, 50 s, 60 s, 70 s, 80 s, 90 s, Ageism, Labourforce, Demographics, Workforce, Workplace, Workers, Retirement, Recruit, Aging, Leadership, Entrepreneur, Preschool Child, Child, Preschool, Infant, Children’	57K

In our previous study, ⁶² we have collected our dataset from LinkedIn (posts) for four months (March 2022 – July 2022) using search query terms shown in Q2, Table 2. These posts are anonymized (aggregated) data, and we do not use personal information. Then, posts are saved in a CSV file, including around 57 K posts. The CSV file includes two columns, the query term and the post.

Data preprocessing

Preprocessing steps, as shown in algorithm 3, involve removing duplicate articles and irrelevant characters, tokenizing, removing stop words and lemmatizing with POS tags. Python's Pandas package is used to read and save the CSV file as a data frame. Redundant data is removed in the second step, leaving around 31 K articles. Third step, we eliminated all unnecessary characters, including several Unicode characters. The fourth step tokenizes the texts using a simple preprocess function from the Python package ‘Gensim’. In the fifth step, stop words are deleted from the articles. The final step involves lemmatization using Spacy. The result of preprocessing is cleaned texts, which are then used in the LDA model.

Algorithm 3 data-pre-processing

Input: CSV_file

Output: processed_data

1: data_df ← read CSV_file

2: removed_dup_df ← remove duplicate data_df

3: removed_irr_char_df ← remove irrelevant characters from removed_dup_df

4: token_df ← tokenize removed_irr_char_df

5: rsw_df ← remove stopwords from token_df

6: lemm_df ← lemmatize rsw_df

7: processed_data← clean lemm_df

Topic modelling using unsupervised machine learning

A topic model is a machine learning technique for discovering topics in a set of documents. ⁶³ Information filtering, retrieval and semantic search are some applications of topic modelling used in document analysis. A topic model can also be referred to as a latent topic model or a latent semantic model. This method analyzes words and their variations in various contexts to discover hidden themes in a text.

We used LDA ³¹ , a topic modelling method, in this study. We set the parameters for our LDA model to 15 topics, 10 passes and 100 iterations. There is no doubt that the number of topics is an important parameter when building a model. When the number of topics is large, the model must be overfitted, but when it is small, it must be underfitted. By creating multiple LDA models with different topics (k) and comparing the coherence measures with the visual representation, we could determine the optimal number of topics. Passes refer to the number of times an algorithm must traverse the whole corpus. The maximum number of iterations is required to determine the probability of each topic in the corpus.

In topic modelling, coherence scores are used to measure the human interpretability of a topic. The topic coherence scores give an extensive method for comparing various topic models. It compares by capturing the optimal number of topics and gives a ‘Coherence Score’ for those topics’ interpretability. Recent research has focused on measuring the coherence of topics to address the issue that topic models do not guarantee the interpretability of their output. The most common methods for adjusting LDA hyperparameters are based on various topic coherence measures. In this work, we used the C_v, C_uci, C_umass and C_npmi measures of coherence. C_v measure uses a sliding window-based one-set segmentation of the most popular or top words and an indirect confirmation measure. The indirect confirmation measure employs cosine similarity and normalized pointwise mutual information (NPMI). C_uci measure employs a sliding window and all word pairs’ pointwise mutual information. C_umass measure is a confirmation measure, and it utilizes co-occurrence counts for documents, a one-preceding segmentation, and a logarithmic conditional probability. C_npmi measure can be considered an enhancement of C_uci coherence, which uses the NPMI.

Parameter discovery & quantitative analysis

We identify the parameters and macro-parameters based on domain knowledge and quantitative analyses utilizing tools such as term scores, intertopic distance maps and word clouds. The keyword scores for each parameter are represented graphically in descending order. This term score visualization substantially influences parameter identification. A two-dimensional intertopic distance map depicts the parameters as parameter circles whose sizes correlate to the number of words in the dictionary that define the parameters. A word cloud is a representation of a group of words. Word Cloud highlights the famous words and phrases in the articles based on how often they appear and how important they are. It gives quick, easy-to-understand graphics that can lead to in-depth analyses.

Validation and visualization

Results can be both internally and externally validated. Internal validation of a parameter means studying and reviewing the relevant papers. In our research, documents may be academic articles or posts. We described how we interpreted the relationship between the papers and the parameters. External validation is performed by comparing the two datasets’ parameters, keywords and metric metrics. For internal and external validation, many visualization methods are utilized for the visualization. Numerous visualization techniques are utilized to describe the datasets, document clusters and the discovered parameters using histograms, taxonomies, Term scores, Intertopic Distance Map and word clouds. Several Python packages, such as LDAvis, Plotly and Matplotlib, are used to construct these visualizations.

Parameter discovery for multi-generational labour markets (web of science)

In this section, we describe the parameters for multi-generational labour markets that were detected using our LDA model based on the Web of Science dataset, including the taxonomy and quantitative analysis. The parameters are divided into five macro-parameters. The five macro-parameters are discussed in Sections 3.1 to 3.5.

Based on the LDA model, 15 parameters were found in the LinkedIn dataset. This 15-parameter set was grouped into five macro-parameters. Based on the coherence score, we determined 15 clusters (which will be discussed in more detail in the next section). Table 3 lists the LinkedIn dataset parameters and macro-parameters. According to Column 1, the parameters are categorized into five macro-parameters: Generations-specific Issues, Learning & Skills, Employment Sectors, Consumer Industries and Employment Issues. Columns two and three list parameters and cluster numbers, respectively. According to the fourth column, each parameter has a certain percentage of keywords. As indicated in the fifth column, each parameter is associated with the top keywords.

Table 3.

Parameters and macro-parameters of multi-generational labour markets.

Macro-parameters	Parameters	No	%	Keywords
Learning & Skills	Digital Learning Environments	3	9	System, Data, Model, Generation, Base, Propose, Method, Using, Network, Approach, Work, Problem, Use, Generate, Application, Result, Time, Paper, Analysis, User.
		11	4.2	Learning, Student, Education, Course, Teaching, Skill, Experience, Language, Base, Study, Group, Learner, Online, Environment, Cohort, Design, Educational, Classroom, Assessment, University.
	Management Science	2	11.1	Social, Community, Experience, Work, Research, Practice, Identity, Interview, Study, Within, Article, Cultural, Explore, Context, Generation, Theory, Qualitative, Challenge, Paper, Diversity.
	Health Sciences	14	3.5	Medical, Training, Study, Skill, Clinical, Program, Result, Method, Survey, Assessment, Practice, Cohort, Evaluation, Professional, Participant, Knowledge, Test, Scale, Group, Question.
	Learning & Employment in Immigrant Families	10	5.1	Student, Education, University, Academic, Higher, Study, College, School, Program, Graduate, Career, Science, First, Research, Year, Experience, Skill, Faculty, Cohort, Degree.
Employment Sectors	Education Sector	4	8.6	Teacher, School, Education, Teaching, Professional, Educational, Science, Development, Primary, Study, Training, Educator, Practice, Secondary, Service, Covid-19, Experience, Classroom, Pandemic, Curriculum.
	Eldercare Sector	8	7	Health, Care, Services, Mental, Police, Service, Need, Patient, Support, Trauma, Older, Life, Work, Disability, Social, Treatment, Experience, People, Population, Caregiver.
Consumer Industries	Marketing & Consumer Behavior	15	2.1	Consumer, Customer, Product, Marketing, Service, Brand, Food, Consumption, Experience, Market, Purchase, Value, Industry, Fashion, Tourism, Price, Environmental, Retail, Housing, Quality.
Learning & Employment Issues	Multi-Generational Workforce	13	3.6	Knowledge, Research, Development, Management, Business, Generation, Design, Process, Innovation, Paper, Approach, Project, Model, Base, Work, Study, Need, Value, Practice, Methodology.
	Employee Satisfaction	7	7.9	Example, Study, Work, Factor, Employee, Satisfaction, Demographic, Relationship, Research, Findings, Result, Performance, Difference, Influence, Gender, Attitude, Variable, Level, Intention, Perception, Analysis.
	Mental Health	9	5.6	Risk, Violence, Stress, Self, Experience, Study, Factor, Abuse, Adolescent, Among, Social, Sexual, Emotional, Behaviour, Relationship, Burnout, Demographic, Health, Report, Associate.
Generations-specific Issues	Technology Usage	1	11.5	Information, Technology, Digital, Medium, Social, Library, Online, User, Communication, Internet, Literacy, Game, Generation, Mobile, Computer, Study, Network, Experience, People, Research.
	Culture & Identity	12	3.9	Generation, Cultural, Culture, Life, Political, Chinese, History, Article, World, Identity, Society, People, Memory, Young, Values, Experience, Story, Narrative, Historical, Language.
	Immigrant Families & Intergenerational Life	6	8.5	Child, Family, Woman, Parent, Cohort, Immigrant, Life, Intergenerational, Study, Mother, Adult, Fertility, Gender, Data, Employment, Experience, Among, Young, Time, Result.
		5	8.5	Country, Economic, Policy, Employment, Market, Change, Demographic, Social, Population, Growth, Migration, Financial, State, Increase, Rural, Labour, Economy, Worker, Migrant, Development.

In Figure 3, we show a taxonomy of the multi-generational workforce. To create the taxonomy, we used the parameters and macro-parameters from the Web of Science dataset. First level branches display macro-parameters, and second level branches display discovered parameters.

Figure 3.

Web of science perspective taxonomy of multi-generational labour market.

Models were created with Gensim LDA package and themes were rendered with PyLDAvis. Based on the coherence measures, we selected different top models with the optimal number of topics. Figure 4 presents four different coherence measures for finding the optimal number of topics. Our choice of 15 topics was since it appeared to be an optimal number. Furthermore, the intertopic distance map has fewer overlapping circles.

Figure 4.

Coherence measures.

Based on the extracted 15 topics information, Figure 5 shows the intertopic distances and the most important words. Topics are represented by circles on the intertopic distance map. Depending on the topic's relevance, the circle size increases significantly. Typically, a good topic model will have fewer overlapping circles throughout the chart. A poor topic model, on the other hand, would have circles that overlap heavily clustered in one quadrant.

Figure 5.

The intertopic distance map of the topics and the most important words.

In Figure 6, the histogram shows how many documents are in each topic. On the x-axis, the number of topics is displayed, while on the y-axis, the number of documents is displayed. As an example, Topic 12 has approximately 3500 documents. This visualization provides insights into the document distribution among the various topics, allowing us to identify the topics that have a higher or lower number of associated documents.

Figure 6.

Number of documents in each topic.

Figure 7 shows the document word count and the number of documents. Most documents in the dataset contain fewer than 500 words. A few contain between 500 and 750 words, indicating a moderate length. Furthermore, even a fewer article abstracts contain between 875 and 1000 words, suggesting a relatively smaller proportion of longer documents in the dataset. This information helps to understand the distribution of document lengths within the dataset and provides insights into the general size and structure of the documents being analyzed.

Figure 7.

Histogram (web of science article abstracts).

Figure 8 shows the most frequently occurring and relevant words and phrases in the articles. A word cloud provides a visual representation of the most prominent terms associated with each topic. Analyzing the size and prominence of the words allows us to gain insight into the key themes and concepts discussed in the articles. In general, the larger and more prominent the word, the more likely it is to appear in the topic. We can use this technique to identify the articles’ main ideas and focus areas without having to read their entire text. Nonetheless, word clouds alone may not provide a comprehensive understanding of the articles, so they should be used in conjunction with other analytical methods.

Figure 8.

Word cloud for each topic.

Figure 9 displays the ten most significant keywords for each parameter. The parameter keywords are represented by vertical lines, while the importance scores are depicted by horizontal lines. The colors used in the figure indicate the level of importance, with light maroon indicating the highest level of significance and dark maroon representing the least important keywords. This visualization allows for a quick assessment of the most critical keywords associated with each parameter; we can identify the key terms that contribute significantly to the analysis. By considering the importance scores and the corresponding keywords, we can gain valuable insights into the factors that drive the analysis and the specific aspects that are most relevant to the parameters being examined.

Figure 9.

Term score.

Learning & skills

The Learning & Skills macro-parameter includes four parameters. The parameter Digital Learning Environments includes two clusters 3 and 11. It includes Next-generation systems which are any system designed to work with new technologies, such as AI ⁶⁴ and the Internet of Things. ⁶⁵ While these systems are often used to improve business processes and make them more efficient, they can also help analyze data, predict trends, automate tasks and be used in learning environments. Digital Learning Environments requirements of Generation Z differ from those of previous generations. Teachers need to know how to use digital tools and spend more time on social media to connect with Gen Z students. Building collaborative E-learning environments ⁶⁶ by using tools that make it easy for the instruction creator to design and construct complete collaborative settings. Also, allowing students to share their knowledge and experiences, this learning paradigm helped to increase the productivity of the online courses. Moreover, giving insights about massive open online courses. ⁶⁷ Also, provides the importance of assessment tools for next generation students learning processes. ⁶⁸ Also, Augmented reality learning to engage youth in urban environmental planning. Engaging the young generation in urban planning is a comprehensive strategy that acknowledges their crucial role in determining the future of urban life. Urban environmental concerns have been shown as untied relationships between communities and the young due to their passions for exploration, engagement with nature and analysis.

The parameter Management Science includes qualitative studies of the social, cultural and personal aspects of society that are, in essence, a study of the interaction between people. It examines the way groups, organizations and individuals interact with each other to form communities and networks and build relationships. This parameter explores human experiences, skills and other aspects and how it has changed over time.^69,70 For instance, Qualitative interviews were conducted to explore the experiences of second-generation Asian Americans seeking employment. ⁷¹

The Healthcare Sciences parameter is characterized by diverse healthcare needs of various generations, promotion of health, education and training, workforce engagement and retention, etc. The quality of healthcare is determined by all members of the healthcare team, not just the individual service providers; an approach to providing patients with the best possible quality of care through continuous assessment, training, evaluation and improvement of the skills among healthcare practitioners. For instance, well-trained patient-centered physicians are known for their fundamental practices of communicating with patients. ⁷² To foster and sustain patient-centered attitudes, it is essential to guarantee job satisfaction and communication skills training. On the other hand, Residency programs ⁷³ can select medical students for interviews and employment using specific metrics. These include USMLE scores, GPA and class rank/quartile. In addition, A validated search assessment instrument can effectively measure improvements in residents’ search skills to demonstrate training efficacy and meet practice-based learning competency requirements of the Accreditation Council for Graduate Medical Education. ⁷⁴ Also, the development of the nursing workforce is crucial to achieving the health-related Millennium Development Goals and would provide better medical care to China's rural population. ⁷⁵

The parameter Learning & Employment in Immigrant Families captures the academic issues of the first, second and third generations of immigrants in the United States at different times. For instance, identifying the factors contributing to the retention of first-generation college students and facilitating this retention. ⁷⁶ Also, the challenges faced by first-generation students that hinder the achievement of their academic goals. ⁷⁷ Moreover, exhorting educational institutions to help students overcome barriers and become successful. This can be done, for example, by utilizing work-related employability training for a cohort of undergraduate business school students. ⁷⁸ Also, demonstrating an incorporated curriculum in employability will likely improve students’ employability decisions. In addition, analyzing the skills related across students of the first, second and third generations in the United States. ⁷⁹ Also, students who are first, second or third-generation students are often seen as less confident or less likely to do well in their education. ⁸⁰

Employment sectors

Employment Sectors macro-parameter includes two parameters. The parameter Education Sector emphasizes the fact that preparing the next generation of teachers and workforce is a global priority. ⁸¹ Technology in schools and the excellent performance of so-called Technology Generation or Generation Z students are not adequate to promote digital competence. Technological and pedagogical instructor abilities are crucial.^82,83 For example, the potential benefits of using video games in the classroom include the ability to promote innovation, enhance critical thinking and improve problem-solving. As their use increases, educators need to think critically about the impact a game-based learning environment can have on their students. ⁸⁴

The parameter Eldercare Sector highlights the importance of eldercare in labour markets. It emphasizes that the meaning of age has altered as Americans live longer, especially seniors and Baby Boomers. Thus, programs are created to increase students’ civic involvement and professional growth, enable community debate on active aging and prepare them for gerontology. ⁸⁵ Moreover, baby boomers’ health may be a significant factor in explaining their current workforce participation, with important implications for health expenditure and support in old age. ⁸⁶ In addition, Gerontological social workers are needed to care for the aging population, according to social work authorities. ⁸⁷ For this purpose, a gerontological social educator must help future, and current practitioners understand how to help their patients live a fulfilling old age. Observing students and practitioners demonstrate competence in carrying out taught skills is essential to ensuring this outcome. ⁸⁸

Consumer industries

Consumer Industries macro-parameter includes one parameter but can be extended using modelling of additional datasets. The Marketing & Consumer Behaviour focuses on how consumers make purchasing decisions. This includes everything from demographics and different generations to attitudes to social media usage.^89–91 For instance, Indian consumers’ definition of retail service quality concerning small retailers. However, it did not consider demographic variables such as income, education and occupation. ⁹² On the other hand, comparing consumers from different generations within a target segment – bank service customers. ⁹³ Also, differences between tourist satisfaction levels according to their demographic characteristics: gender, age, educational level and employment status. ⁹⁴ Moreover, the issue of the growing number of elderly customers, baby boomers who retire, become more demanding of high-quality goods and services. However, they also have more difficulty adjusting to the store environment and shopping. ⁹⁵

Learning & employment issues

Employment Issues macro-parameter includes three parameters. The parameter Multi-Generational Workforce describes the workplace environments across different generations. Multi-generational workforce may be vital to creating an adequate knowledge working environment and contributing to the goal of organizations being more productive and competitive. ⁹⁶ For example, Highlighting the values and work ethic of Generation Z while briefly describing the characteristics of the Baby Boomers, Generation X, Generation Y and Generation Z. ⁹⁷ The labour market is examined by studying the behaviour of Generation X and Generation Y (Millennials). ⁹⁸ In the foreseeable future, it is anticipated that these two generations will significantly shape the labour market. On the other hand, the primary abilities and knowledge of a new young generation of Library Information Services (LIS) workers has been discussed. ⁹⁹ Also, an overview of the digital world and its effects on LIS workers has been provided. Moreover, research on age-friendly workplaces focuses on the physical work environment, positive organizational attitudes, and the link between employee experiences and performance. ¹⁰⁰

In the Employee Satisfaction parameter, research showed different characteristics associated with each generational cohort in the workplace. Generation Y and Z are far more likely to be competent in technical fields. There is no doubt that generation X is technically behind gen Y. To increase job satisfaction, we must understand how challenges, issues and characteristics work. ¹⁰¹ There was a difference between Baby Boomers and Generation X workers in job satisfaction, organizational commitment and willingness to leave the workplace. According to the results, boomers are more likely to be satisfied at work and are less likely to quit than Xers. ¹⁰² Moreover, statistics showed that job satisfaction positively correlates with age. ¹⁰³

The parameter Mental Health stresses the increasing mental health problems related to multi-generational labour markets. Many people and especially young ones, feel overshadowed and the feeling of being constantly watched and judged. They also lack experience with bullying, family dysfunction or emotional abuse, making them vulnerable to depression. ¹⁰⁴ All these things can cause strain on their self-esteem, making it more likely for them to develop depression symptoms. In addition, they have new responsibilities such as making money, finding jobs and school; increased stress from all of that; and often less support from family members who are moving on with their lives or are not around at all. ¹⁰⁵ This can lead to feelings of isolation and helplessness, a particular concern in older people. ¹⁰⁶ In addition, workplace bullying is a form of workplace mobbing, the negative behaviour exhibited by coworkers that can cause severe psychological distress. ¹⁰⁷ In many countries, it is illegal for an employee to bully others at work – including via electronic means such as social networking sites. On the other hand, smartphone addiction has been widely studied in recent years. Smartphones could affect various demographic, personality-linked psychological and emotional variables. ¹⁰⁸ Thus, the study to understand the elements that influence smartphone addiction across generations may lead to more effective education, productivity in the workplace and awareness efforts on technology's impact on mental health. Also, smartphone users from Generations X, Y and Z workers answered an 80-item questionnaire. ¹⁰⁹ Anxiety over smartphone use was measured by characteristics such as social pressure and emotional gain. The study's key finding is that Generation Y has much more addictive behaviour than the other two generations. The most significant predictors for all three generations were social environment pressure and emotional gain. Interestingly, emotional gain from smartphone use was more significant for generation Z than for older generations. Like neuroticism and everyday usage time, WhatsApp usage was a significant predictor for Generation Z.

Generation-specific issues

Generations-specific Issues macro-parameter has three parameters. The Technology Usage parameter talks about the digital age,^110–113 which has changed how we communicate, learn, work and consume information. It also discussed the challenges for people who are not digitally savvy, such as older people (baby boomers). For instance, Mobile phone penetration has changed the traditional dynamics of older people leading. Now they depend upon young people such as Gen Y and Z to meditate to help them use mobile phones. ¹¹⁴ Positive interactions increase the provision of care by young people to older people. Moreover, it can utilize smartphones and other technologies to enhance the learning outcomes of Gen Z students who are already proficient in using such technologies. ¹¹⁵ Also, in book preferences, the preference for print and electronic books for reading purposes has many factors such as gender, age, education level, income, ethnicity, etc. For example, Baby boomers and Gen x are more likely to prefer paper books over electronic ones. ¹¹⁶ Also, examining smartphone use categories, process and social, for connections with depression and anxiety and problematic smartphone use. ¹¹⁷ In addition, examining the relationship between process and social smartphone usage, emotional intelligence, social stress, self-regulation, gender and age. For instance, age influences process, social usage and stress. Older people are less likely to acquire smartphone addiction. ¹¹⁸

The parameter Culture & Identity captures multi-generational workforce issues relevant to it. There are many different cultures in the world, each with its unique history and heritage. In order to understand and explain our current identity in our culture, the work of previous generations can be studied.^119,120 For instance, the fundamental differences between Generation X and Baby Boomers regarding cultural traits and work ethics. ¹²¹ On the other hand, the current generation enters a discourse with history and decides what needs to be kept or added and what needs to be discarded. For example, Chinese co-ethnics from the first world who work in the People's Republic of China may uniquely use their Western culture and schooling and are presumed to understand Chinese culture. ¹²²

The Immigrant Families & Intergenerational Life parameter includes research about the issues faced by immigrant families who come to live in the United States or any other country from their original one.^123–125 For example, A comparative study is conducted between women and their descendants, looking at whether family formation differentially affects their labour market position. Regarding employment levels, migrant women of first and second generations have higher unemployment rates than native women, regardless of background characteristics, suggesting differential employment opportunities and family policies among these groups. ¹²⁶ Also, the intergenerational mobility pathways across migrant and non-migrant families in the education and occupational domains are examined. ¹²⁷ On the other hand, the solidarity of a family and a healthy intergenerational relationship between a grandparent and grandchild is important. ¹²⁸

Parameter discovery for multi-generational labour markets (LinkedIn)

Twitter and Facebook have been the subject of intensive research over the past few years, and datasets are readily available. LinkedIn is the largest network of professionals and hence it is an exciting source of data to explore. However, LinkedIn does not have a dataset gathering information (posts) and hence it is challenging to develop datasets from it. LinkedIn was our data source to conduct a study focusing primarily on labour markets involving professionals. A web scraping and social mining technique was used to obtain the dataset. This section presents the parameters detected using our LDA model based on the LinkedIn dataset. We briefly describe in this section the parameters and macro-parameters discovered from the LinkedIn data and use them to develop the multi-perspective taxonomy of multi-generational labour markets. A detailed description of the parameters can be found in.^62,129

We detected 15 parameters. Two of them were merged into one. One of them was discarded since it was in a different language. So, we are left with 13 parameters. These parameters are divided into five macro-parameters. The parameters are categorized into five macro-parameters: Learning & Skills, Business & Employment Sectors, Consumer Industries and Learning & Employment Issues and Generations-specific Issues macro-parameter has one parameter, which is Crimes & Racism. Learning & Skills macro-parameter has two parameters: Learning & Skills and Leadership. Business & Employment Sectors macro-parameter has four parameters: Remote Work, Recruitment, Entrepreneurship and Family Business. Consumer Industries have six parameters: Brand Marketing, Retirement, Energy Sectors, Entertainment, Celebrations and Mental Health. A taxonomy of the multi-generational workforce is shown in Figure 10.

Figure 10.

LinkedIn perspective taxonomy of multi-generational labour market.

The parameters addressed numerous issues in the labour market. The millennial generation, for instance, should understand the challenges associated with working from home and how to overcome them despite the benefits remote work offers. A significant challenge of working from home is the increased risk of burnout. Office environments give employees time to decompress and recharge after a long day at work. Employees who work from home are not afforded this luxury, so they are often under more pressure to be productive constantly. As a result, burnout can occur, and high-stress levels can negatively impact health. To overcome that, it is essential to take breaks throughout the day. This way, employees will be energized and ready to tackle another productive workday. Additionally, mental health needs to be taken care of because it can affect physical health and relationships. To feel comfortable speaking with friends and family members about their feelings, seniors and adults should feel free to express themselves.

Our results also touched on hiring and retaining employees, which is becoming more challenging for organizations. In addition to managing their workforce effectively, organizations have difficulty hiring and retaining the right employees. Depending on the organization, multi-generational employees may leave due to a lack of advancement opportunities. Additionally, creating a culture that appeals to all generations can be challenging. As a result, organizations may need to adjust their policies to be more inclusive of different generations. In addition, flexible work schedules, relevant training programs and open communication between generations can enhance employee satisfaction.

In addition, we found that the number of retirees in the United States will outnumber children by 2023. The workforce will be significantly affected by this change. There will be a decreased labour pool, resulting in higher employee demands and increased costs of hiring, training and finding new employees. Offering part-time schedules and flexible scheduling options to retirees is one solution.

In addition, we found that unemployed Generation Z and Y youth are five times more likely to have criminal records. Furthermore, if racial issues occur at work, they can severely affect both the victim and the workplace. There is a possibility that victims will feel uncomfortable or unwelcome in their jobs, and they may lose respect for colleagues and co-workers. Employees can also distrust each other in a racist environment, which hinders teamwork. Employees with racist attitudes are also more likely to be unproductive and leave the company more often.

Despite challenges in the 21st-century economy, businesses are in an excellent position to remain competitive. The reason for this is that entrepreneurs can be of any age. An entrepreneur can start a small business at any age, or an employee at an established company can start a business. Also, they can consider investing in startups or engaging in other entrepreneurial activities. Although entrepreneurs will have many challenges ahead, they must be prepared for them. Additionally, from the perspective of the brand marketing of the business, it was found that modern digital technology and media are required to meet the needs of youth customers (Gen Y and Gen Z).

Discussion

In this paper, we proposed a data-driven AI based approach to automatically discover parameters for multi-generational labour markets using academic literature and social media analysis. Specifically, we discover parameters from Web of Science and LinkedIn posts using the LDA algorithm. We developed a software tool from scratch for this work that implements a complete machine learning pipeline using two datasets.

We discovered 15 parameters within the Web of Science dataset and categorized them into five macro-parameters: Learning & Skills, Employment Sectors, Consumer Industries, Learning & Employment Issues and Generations-specific Issues. Figure 11 shows the word cloud of keywords discovered from Web of Science articles, with the size of each keyword denoting its frequency, a measure of its importance. According to the figure, the Web of Science is primarily concerned with work, education, experience, research and generation.

Figure 11.

A word cloud generated from web of science.

Using the LinkedIn data, a total of 13 parameters were discovered and categorized into five macro-parameters. The macro-parameters are the same as for the Web of Science, however, there are differences in their constituent parameters. Figure 12 shows the word cloud of keywords discovered from LinkedIn posts, with the size of each keyword denoting its frequency, a measure of its importance. According to the figure, LinkedIn is primarily concerned with work, education, research and generation. These foci though are somewhat similar to the Web of Science the overall mix of foci is different.

Figure 12.

A word cloud generated from LinkedIn posts.

Figure 13 shows a multi-perspective (academic and professionals) view of the generation characteristics in labour markets discovered by our tool. We depicted the importance of hiring and retaining employees. In order to hire and have the right employees, organizations may need help managing their workforces. Several generations may leave the workplace due to a lack of advancement opportunities. Creating a culture that appeals to all generations is also difficult. This issue may require organizations to make their policies more inclusive of different generations. Among the things that can be done are offering relevant training, offering flexible work schedules and establishing an open line of communication between generations.

Figure 13.

A multi-perspective (academic and professionals) view of the generation characteristics in labour markets.

Also, we found that baby boomers in their 50s and 60s are retiring, and more retirees will live in the US in 2023 than children. As a result, the workforce will be significantly affected. With a shrinking labor pool, hiring and training new employees will become more expensive as there are fewer qualified applicants available. A possible solution is to hire more retirees and offer them part-time work.

Moreover, unemployed American youth (Gen Y and Gen Z) have five times more criminal records than employed youth. Moreover, racism at work can harm both the victim and the workplace. As a result of abuse, victims may feel unwelcome at work and lose respect for their coworkers. In addition to discouraging teamwork, racist workplaces can also foster distrust. It is also possible for racist attitudes to decrease employee productivity and turnover.

On the other hand, Today's businesses are well-positioned for success in the 21st century. It is possible to start a small business or to work for an established company at any age since entrepreneurs can start businesses at any age. Other entrepreneurial activities include investing in startups or participating in other entrepreneurial ventures.

Multi-generation workforce skills and benefits

We discuss here the skills and benefits of multi-generational workforce derived from the discovery and analysis of parameters conducted in this study (consider also the multi-perspective view depicted in Figure 1). In the past decade, technology has profoundly impacted the workplace. It has transformed how we work and interact with one another, how we manage our time and how we complete tasks. ¹³⁰ This transformation is particularly evident when examining how technology has changed how employees perform their jobs. These changes are especially apparent when examining how multi-generational workplaces have evolved and continue to devolve into a new normal. ¹³¹ In today's multi-generational workforce, different attitudes and experiences among baby boomers, Gen Xers, Millennials (Generation Y) and Gen Z present greater variety in adopting and using technology. ¹³² With more generations working together, we see wider differences in how various age groups view technology, processes and workplace productivity. Younger generations may share technology skills, while older generations impart sectoral knowledge and general wisdom in the workplace. ²⁵ The millennial generation is now entering the workplace in droves,^130,133 bringing their distinct Millennial values with them as they take on leadership roles and manage other employees who may be from an older generation or younger generation than themselves. As it stands today, millennials are the largest generation currently in the workforce. ¹³⁴

On the other hand, the shift in technology and paradigm among employers also reflects employee trends. The email and phone poll responses seemed unanimous; these trends are not going away anytime soon because technology is being used to help solve problems in the workplace and develop growth opportunities within companies from all sectors. Employees from different generations can now work collaboratively from remote locations and meet virtually without ever having met face to face throughout history, with systems like video conferencing furthering their capability to interact virtually for remote meetings.

Currently, employees no longer retire at 65 or remain with the same organization for decades. This suggests that the days when employees used to work their way up the ladder are long gone. With such a diverse talent pool, businesses can garner a competitive edge through the implementation of a multi-generational workforce. With the workforce becoming increasingly diverse, it is crucial to comprehend the advantages of an inter-generational workplace. Conceptually, an inter-generational workforce is one that has employees from various age groups, including Baby Boomers, Generation X, Millennials and Generation Z. A multi-generational workforce is defined as employees made of individuals emanating from diverse generations. This section focuses on the definition, benefits and challenges of a multi-generational workforce.

A multi-generational workforce provides an organization with a plethora of benefits. The first advantage identified in the literature is that it can result in more effective communication since employees of diverse age groups have different ways of communication. For example, Generation Z and Millennials may prefer communicating through messages or social media whereas the Generation X and Boomers may prefer face-to-face communication or emails. ¹³³ Organizations can tap into the several styles of communication that the different age groups use to disseminate messages more effectively. Because of the diversity of the workforce, organizations can grab this as an opportunity for employees to gain insight from each other and enhance communication across the various generations. A multi-generational workforce can result in better communication due to the various communication styles of the different generations. Understanding and respecting such differences can be critical for bridging the gap and enhancing understanding between employees. ¹³⁵ A cohesive working environment that such a workforce has can be beneficial in improving the overall productivity of the organization.

The second benefit of an intergenerational workforce is that it can offer various experiences and skills. Employees of various ages usually have varying experiences and skills to offer. For instance, Millennials and Gen Z may have more familiarity with new technologies, whereas Gen X may be experts in traditional methods. ¹³⁶ Mixing the various age groups within the organization can enable them to benefit from various experiences and skills. The other advantage is better creativity and innovation for meeting the dynamic marketplace needs. Different age group employees bring various experiences and perspectives on things such as product development, work processes or customer service, thus resulting in innovative and exciting solutions to issues when companies tap on them. ¹³⁷ An intergenerational workforce can also assist in creating an innovation culture since employees feel free to share their ideas. In this regard, businesses which adopt a multi-generational workforce have more likelihood of being successful in the current dynamic marketplace since they have better innovation and creativity that guarantees them a competitive advantage.

Better adaptability and flexibility have also been cited as a critical benefit since employees of various age groups often have varying skill sets. For instance, Millennials and Gen Z are more tech-savvy and embrace new technologies quickly, ¹³⁸ whereas older generations such as Generation X can have more institutional knowledge and experience. Having a multi-generational workforce can ensure that the organization becomes more adaptable to change and flexible. Moreover, a multi-generational workforce can result in better problem-solving. With a combination of backgrounds and ages, employees can introduce various perspectives for solving workplace problems, thus resulting in more effective and innovative solutions. The more well-rounded team that emanates from a multi-generational workforce can enable an organization to tap into a wider pool of knowledge and skills. It can make the team to become more versatile and meet the workplace challenges. ¹³⁹ Younger workers can gain insights from experiences of older workers, thus help in creating a more supportive and collaborative work environment. By creating a dynamic workplace, a multi-generational workforce increases the likelihood for more exchange of ideas and vibrant work culture. Increased profitability and productivity are the other benefits of a multi-generational workforce. Employees of different backgrounds and ages can introduce various viewpoints on workplace issues. ¹⁴⁰ The more potential for exchange of ideas among the different backgrounds and ages can result in better profitability and productivity within the workplace.

Multi-generation workforce challenges

We discuss here the challenges that a multi-generational workforce poses. These are derived from the discovery and analysis of parameters conducted in this study.

Notwithstanding the numerous benefits accrued from a multi-generational workforce, it also poses certain challenges. One of the major challenges is the management of various expectations on work–life balance. For instance, Generation Z value work–life balance more than Generation X. In this regard, they are likelier to leverage on flexible work arrangements such as compressed workweeks or telecommuting. ¹⁴¹ As a generation that directly encountered the great recession, Generation Z are more concerned about job security, perks and salary, but at the same time tend to be vocal regarding work–life balance as well as workplace flexibility. Notably, work–life balance has been found to be quite pertinent not only for Generation Z but all other employees. ¹⁴¹ In fact, this generation holds the assumption that it is the mandate of the organization to offer flexibility since it only increases efficiency and productivity, ¹⁴² and that such a work arrangement needs to be open. Balancing such interests with the other demands of other generations regarding work-life balance may be quite challenging for an organization.

Communication is an important tenet for the workforce. However, different generations have different communication styles. Whereas Gen Xers often prefer emails and calls, Generation Y often prefers to send instant messages. The use of informal language, abbreviations and colloquialisms further contributes to communication breakdown, thus making communication a major challenge. To address this, organizations should encourage collaboration among the different generations to ensure they learn from each other. ¹⁴³ Changing the collective mindset and influencing different generations to view themselves as partners can ensure everyone benefits from new communication forms and ideas. Motivating a multi-generational workforce is also quite daunting. Motivating employees usually implies the creation of a corporate culture that supports all persons’ goals and ideals. It can include flexible working strategies and perks that enable various generations to pursue diverse aims within their careers and work in several ways to actualize them. To address this, it is crucial for multi-generational workplaces to treat each employee as an individual. Instead of motivating all persons with similar benefits, an organization should personalize its motivational approach. ¹⁴³

Every generation has certain stereotypes which may be harmful to the workplace. For example, older workers like Gen X often believe that Millennials and Gen Z are entitled and tech-obsessed, whereas the younger employees such as Gen Z believe that Baby boomers and Gen Z are stubborn and old-fashioned. ¹⁴⁴ This illustrates that while various generations embrace different styles and preferences of work, stereotypes can negatively affect the work environment. To tackle this, it is crucial to concentrate on valuing employees for their individual strengths. ¹⁴³ Employers should not assume that certain individuals in the team require special treatment and help. They should not concentrate on weaknesses of other workforce members. Rather, they need to know each employee individually, and concentrate on tapping on their strengths.

Furthermore, it is difficult to balance the weaknesses and strengths of a multi-generational workforce. Every generation has its characteristics that it delivers to the workforce. Such differences need to be embraced to enable organizations leverage on their teams. However, managers at times view the gaps that exist between members of teams as negatives. Despite this, developing a team that brims with diverse perspectives and insights can be beneficial to an organization. ¹³² To tackle this, cross-generational mentoring can play a pivotal role, and this can be done through the development of a reciprocal mentoring program. Such a program can ensure, for example, that Millennials teach Gen X how to utilize social media along other forms of technology, whereas Baby Boomers can guide Gen Z into communication and interpersonal skills while sharing knowledge regarding the way the business functions. The organization should allow team members to gain insights from each other and depend on other organizational members when they need assistance to balance their weaknesses and strengths. In addition, the organization should fine-tune onboarding programs to meet the needs of the various generations, especially Generation Z. ¹⁴² A well-designed program can enable new hires to minimize uncertainty and anxiety and provide knowledge and clarity to their role. Effective onboarding can also lead to better job satisfaction, loyalty and performance. Mentoring can be instrumental in ensuring the younger employees coordinate with various organizational processes.

Conclusions

Capitalist approaches have had negative impacts on social, economic and environmental sustainability. Economic issues such as inflation and energy costs have been exacerbated by global events and financial crises have revealed weaknesses in modern economies. The current trend of people quitting their jobs in large numbers, known as the Great Attrition, and the presence of multiple generations in the workforce also pose challenges. Transformative approaches are necessary to address these issues and protect society, the economy and the environment.

This study uses big data and machine learning to identify multi-perspective parameters for multi-generational labour markets. The parameters were discovered using 35,000 academic articles and 57,000 LinkedIn posts, and were organized into five macro-parameters: learning and skills, employment sectors, consumer industries, learning and employment issues and generation-specific issues. The study also includes a knowledge structure and literature review of multi-generational labour markets based on over 100 research articles. A machine learning software tool was developed for data-driven parameter discovery and various quantitative and visualization methods were applied to explore the topic.

This paper presents an approach for obtaining comprehensive, objective and multi-perspective information on a subject using machine learning and deep learning and provides tools and resources for accessing information from various datasets. The research in this paper contributes to our understanding of age dynamics in labour markets and may be used to raise awareness and drive future research on the topic using advanced technologies. The findings and knowledge gained from this work can be used to inform decisions and guide labour economics research, and the work is expected to enhance the theory and practice of AI-based methods for information and parameter discovery, extend the use of LinkedIn and scientific literature media for information discovery and promote novel approaches to labour economics and labour markets. Ultimately, this work aims to contribute to the development of sustainable societies and economies.

Artificial intelligence is enabling autonomous functionality in various systems, including self-driving cars and robots, and will be extended to larger systems such as industrial sectors and governance. By identifying and defining the key characteristics and variables that will guide the design and operations, designers and managers can ensure that they are creating a product, system or process that meets the desired specifications and requirements. The same concept is applicable, though much more complex and grander, to the design and governance of economies and societies. Internet of Things, AI, big data and high-performance computing technologies will allow increasing levels of autonomy in economic, social and other governance systems. Discovering system parameters, even in the absence of autonomous capabilities, is necessary for decision-making and problem-solving during design and operation.

The work's limitations include reliance on specific data sources, namely article abstracts and LinkedIn posts, which potentially introduces bias and limits generalizability. The timeframes of the datasets might overlook certain developments and long-term trends in the labour market. While the sample size is large, it may not fully represent the diverse nature of multi-generational labour markets. Additionally, the lack of qualitative data and overlooked variables may restrict a comprehensive understanding of labour market dynamics. Addressing these limitations will enhance the study's relevance and broader applicability in promoting sustainable societies and economies, shaping the direction of future research.

This paper is part of our broader work on the use of information and communication technology (ICT) to address challenges facing smart cities and societies. Our work on this topic has included the concept of Deep Journalism,^145,146 as well as research on topics such as transportation, ¹⁴⁶ tourism, ¹⁴⁷ smart families and homes, ¹⁴⁸ healthcare services for cancer, ⁵⁸ mental health, ¹⁴⁹ education during the COVID-19 pandemic, ¹⁵⁰ energy systems ¹⁵¹ and AI-based event detection. ¹⁵² Future work will be directed to improving the methodological approach presented in this paper using advanced deep learning methods and their applications to investigate and improve labour economics and other problems facing our societies.