A semi-supervised framework for generating multi-dimensional taxonomies from asset maintenance documents
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 operation and maintenance of buildings generate large volumes of unstructured textual data, such as inspection reports and service requests. These records contain valuable insights that can support fault detection, cost tracking, and resource planning. However, existing classification approaches often rely on static, expert-defined labels that fail to reflect the complexity of real-world maintenance operations. This paper introduces a hybrid framework that combines sentence embedding, clustering, topic modeling, and network modularization to uncover recurring patterns in maintenance text. The extracted patterns are then reviewed and refined by facility management experts to develop a multi-dimensional taxonomy model tailored to operational needs. The methodology is applied to a case study involving over 30,000 work orders. The results demonstrate how the proposed system captures fine-grained details such as system type, failure mode, and required trade expertise. A proof-of-concept software tool, developed in collaboration with facility managers, showcases the practical value of the taxonomy in enabling data-driven decision-making, such as identifying cost drivers and recurring issues. Additionally, the resulting taxonomy models serve as effective prompts for zero-shot text classification, enabling large language models to classify new maintenance records without requiring retraining or labeled data. This approach provides a scalable and adaptable foundation for text classification systems in asset management.
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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 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