Bibliographic record
Abstract
Student mobility is a priority in the European Union since it not only allows academic interchange but also fosters the awareness of being a European citizen amongst students. The Bologna Process aimed at homogenizing the structure of the European Universities to facilitate the recognition of academic titles as foreseen by the Lisbon Recognition Convention and student mobility during their matriculation. Over one and a half million students have already benefited from mobility programs such as the Erasmus programme. Students that participate in a mobility program must consider a destination, a selection of courses to follow abroad and how their home institution will recognize their foreign credits. Selecting the most appropriate courses is not a simple task since a course title doesn't always reflect its content. As a result, manual inspection of syllabi is necessary. This makes the task time-consuming since it might require manual inspection and comparison of many syllabi from different institutions. It would be nice to be able to at least partially automate the process -- i.e. given a set of syllabi from two different universities, to be able to automatically find the best match among courses in the two institutions. We started experimenting with this possibility, and although we do not yet have final results we will present the main idea of our project. Our plan is to try to apply similarity matching algorithms to available documents. Similarity matching is often based on co-occurrence of common words. However, a naïve application of such an algorithm would probably end up generating spurious similarities from the co-occurrence of general terms like "hour, exercise, exam...". Using a stop-word strategy in which these words are catalogued and ignored might seem a viable solution, but generally does not significantly improve the results: words that may be considered irrelevant in one context might be important in a different context. The path we are following is to assume the existence of a reference ontology, where all terms have a description, and then try to identify the occurrence of the concepts existing in the ontology within the examined documents. In this way we will be able to state that "syllabus x deals with topic y". The matching between different syllabi would then be calculated by matching the topics that were associated with the syllabi. We decided to focus on the Computer Science domain since the domain has already been classified into areas, units and topics present in CC2001[1] and this ontology has already been mapped into XML structures[2]. We then used a similarity matching algorithm that uses Wikipedia as a reference corpus[3]. Although preliminary results are not yet fully satisfactory, we believe that this might result from working at the word level rather than at a concept level; "software engineering" is not just the co-occurrence of "software" and "engineering" but a more complex concept. We are therefore currently exploring the possibility of identifying multi-words as concepts (still by using Wikipedia as a reference to decide if this is the case or not). If our attempts are successful, the next step will be to (semi-)automatically crawl academic sites to identify curricula and automatically match them by using our algorithm.
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 | 0.002 | 0.016 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.013 | 0.008 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.005 | 0.004 |
| Open science | 0.004 | 0.004 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.029 | 0.035 |
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".