Bibliographic record
Abstract
In this work we employ heuristic search to obtain macro operators for spaces defined in our production system. A macro operator is a sequence of original operators which reaches a subgoal from a state without search. A macro table has operators for each subgoal. Korf (Korf 1985) used macro operators to find suboptimal solutions for the Rubik's Cube and the 15-Puzzle. While the paths found by the macro method are not guaranteed to be optimal, once the macro table is calculated the search effort is negligible. Traditionally macro operators were found by uninformed search methods, because there were no obvious heuristics. We have devised a simple notation, PSVN (Hernadvolgyi & Holte 1999), to represent state spaces. In PSVN, states are vectors of labels and the operators are simple rewriting rules. For this notation we invented a technique to automatically generate admissible and monotonic heuristics to guide the A* family of algorithms. We apply a simple transformation – domain abstraction – on the description of the original space to obtain the abstract space where the distance between two states in the original space is never shorter than the distance between their images in the abstract space. The heuristic values are the lengths of shortest paths in the abstract space. We calculate the distance between the image of the goal state and the rest of the abstract states and store them in a look-up table indexed by the abstract states. This look-up table is also called a pattern database and the method was first used by Culberson and Schaeffer to solve the 15-Puzzle (Culberson & Schaeffer 1994). Korf used pattern databases to solve random instances of the Rubik's Cube for the first time. So far pattern databases have only been used to obtain shortest paths. In this work we use them to find macro operators. A macro operator reaches a subgoal state without search. To solve for a goal state, each macro brings one (or a few) of the labels in the vector representing the state to the index where they occur in the goal state. Subsequent application of the macros in the order of the subgoals fixes all labels and the goal state is reached. The subgoals are patterns where labels at specific indices are identical to the labels of the goal state at those indices. The first subgoal is to fix the label at index i1, the next subgoal is to fix the label at index i2 such that the label at index i1 remains intact and the last subgoal is to fix the last label leaving already fixed labels undisturbed. It
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 | 0.003 | 0.012 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.004 | 0.004 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.005 | 0.009 |
| Open science | 0.003 | 0.004 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.010 | 0.004 |
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".