Extended Abstract

Introduction

The labor market shapes the trajectory of growth, productivity, and social welfare. Central to its functioning is the alignment between the supply of labor and the demand for skills. Yet, this alignment is rarely seamless. In many developing countries, including Iran, education systems often fail to equip graduates with the competencies required by employers. This mismatch, known widely as the “skills gap,” limits economic opportunities, constrains productivity, and undermines competitiveness. Globally, researchers and policymakers have increasingly turned to new sources of data to understand skill demand. Traditional datasets such as household surveys or labor force censuses, while valuable, are limited in scope and updated infrequently. In contrast, online job advertisements provide a dynamic and timely reflection of labor market needs. The study contributes in two major ways. Substantively, it highlights the growing significance of technical and social-emotional skills in Iran’s labor market. Methodologically, it demonstrates how semi-structured digital data can be transformed into actionable insights using text-mining tools. These findings not only enrich academic debates on labor economics and human capital theory but also provide concrete recommendations for policymakers and educational planners to address skill mismatches and improve employability.

 Methodology and Data

The research employs text mining as the primary methodological approach. Text mining, positioned at the intersection of natural language processing, machine learning, and statistics, enables the systematic conversion of unstructured textual information into structured, analyzable data. It is particularly well suited for large-scale analysis of online job postings, which are semi-structured, dynamic, and often contain rich information about required skills, qualifications, and conditions. The dataset consists of around 10,000 job advertisements gathered from Telegram job-seeking channels between 2018 and 2025. Telegram was chosen due to its widespread use in Iran and the diversity of its job postings, which range from professional occupations requiring advanced degrees to entry-level service and manual labor positions. Channels with the largest follower bases were selected to ensure broader coverage. Using the Madeline Proto web service, the data were extracted in JSON format and stored in a MySQL database.

 Data Cleaning and Pre-processing

The raw dataset contained a significant proportion of irrelevant or duplicate content. A filtering process was applied to exclude non-recruitment posts such as advertisements, promotions, or unrelated announcements. After cleaning, the text-mining process proceeded through four major steps: 1) Breaking text into words; 2) Eliminating irrelevant words; 3) Keyword extraction and tagging – identifying skills, occupations, and industry terms using dictionaries built iteratively from the data; 4) Classification – assigning postings into categories such as required skills, industry sectors, occupations, gender preferences, age restrictions, and experience requirements.

 Analytical Framework

For consistency, the study adopted the World Bank’s (2016) skill typology, which classifies skills into four groups: 1) Technical skills (e.g., computer literacy, software use); 2) Advanced cognitive skills (e.g., problem-solving, data analysis); 3) Basic cognitive skills (e.g., literacy); 4) Social-emotional skills (e.g., teamwork, adaptability). Frequency distributions and word clouds were employed to highlight the most in-demand skills and occupations. Additional filters allowed examination of employer preferences regarding gender, age, and experience.

 Findings

Data analysis revealed several key patterns in Iran’s labor market during the study period.

Skills in Demand

Two categories of skills dominated employer requirements: 1) Technical Skills: Advanced technical competencies, prior work experience, and computer literacy were among the most frequently cited in job postings. Within this category, educational qualifications and IT knowledge also ranked among the top ten most demanded skills; 2) Soft and Social Skills: Skills such as negotiation ability, proactivity, attention to detail, professional appearance, professionalism, and teamwork appeared consistently across job advertisements. These findings indicate that Iranian employers, much like their global counterparts, increasingly value social-emotional capacities that complement technical expertise.

Occupations

The five most frequently advertised occupations were: 1) Education-related roles; 2) Sales personnel; 3) Consultants (educational and financial); 4) Marketing specialists; 4) Accountants.

Discussion and Conclusion

This study demonstrates the potential of big data from social media platforms for labor market analysis in Iran, where conventional data sources may be outdated or limited. By applying text-mining techniques to a large dataset of Telegram job postings, the research provides timely insights into the skills, occupations, and demographic preferences shaping demand. The findings confirm the rising importance of soft skills alongside technical competencies. This aligns with international evidence (e.g., Deming 2017; Rios et al. 2020) highlighting the value of social interaction, adaptability, and teamwork in modern labor markets. For Iran, this suggests an urgent need to integrate soft-skill training into curricula, from schools to higher education, to enhance employability. Moreover, the analysis underscores the gap between academia and industry. Employers’ emphasis on practical competencies over formal qualifications reveals the limitations of current educational curricula. Policies should promote stronger linkages between universities and industries, including internships, joint projects, and curriculum reforms tailored to labor market needs. Finally, this research illustrates the feasibility and value of using semi-structured data from social media for labor market intelligence. While challenges remain, such data provide real-time, granular insights that complement traditional statistics. Future work should expand datasets, combine multiple online sources, and refine analytical tools for even deeper understanding.