Combining Techniques to Refine Item to Skills Q-Matrices with a Partition Tree.
Bibliographic record
Abstract
The problem of mapping items to skills is gaining interest with the emergence of recent techniques that can use data for both defining this mapping, and for refining mappings given by experts. We investigate the problem of refining mapping from an expert by combining the output of different techniques. The combination is based on a partition tree that combines the suggested refinements of three known techniques from the literature. Each technique is given as input a Q-matrix, that maps items to skills, and student test outcome data, and outputs a modified Q-matrix that constitutes suggested improvements. We test the accuracy of the partition tree combination techniques over both synthetic and real data. The results over synthetic data show a high improvement over the best single technique with a 86% error reduction on average for four different Q-matrices. For real data, the error reduction is 42%. In addition to the substantial error reduction, the partition tree refinements provide a much more stable performance than any single technique. These results suggest that the partition tree is a valuable refinement combination approach that can effectively take advantage of the complementarity of the Q-matrix refinement techniques. It brings the goal of using a data driven approach to refine the item to skill mapping closer to real applications, although challenges remain and are discussed.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.006 | 0.036 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".