Public Opinion Mining on Construction Health and Safety: Latent Dirichlet Allocation Approach
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The construction industry has been experiencing many occupational accidents as working on construction sites is dangerous. To reduce the likelihood of accidents, construction companies share the latest construction health and safety news and information on social media. While research studies in recent years have explored the perceptions towards these companies’ social media pages, there are no big data analytic studies conducted on Instagram about construction health and safety. This study aims to consolidate public perceptions of construction health and safety by analyzing Instagram posts. The study adopted a big data analytics approach involving visual, content, user, and sentiment analyses of Instagram posts (n = 17,835). The study adopted the Latent Dirichlet Allocation, a kind of machine learning approach for generative probabilistic topic extraction, and the five most mentioned topics were: (a) training service, (b) team management, (c) training organization, (d) workers’ work and family, and (e) users’ action. Besides, the Jaccard coefficient co-occurrence cluster analysis revealed: (a) the most mentioned collocations were ‘construction safety week’, ‘safety first’, and ‘construction team’, (b) the largest clusters were ‘safety training’, ‘occupational health and safety administration’, and ‘health and safety environment’, (c) the most active users were ‘Parallel Consultancy Ltd.’, ‘Pike Consulting Group’, and ‘Global Training Canada’, and (d) positive sentiment accounted for an overwhelming figure of 85%. The findings inform the industry on public perceptions that help create awareness and develop preventative measures for increased health and safety and decreased incidents.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.002 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it