Designing Formulae for Ranking Search Results: Mixed Methods Evaluation Study
Bibliographic record
Abstract
BACKGROUND: A major factor in the success of any search engine is the relevance of the search results; a tool should sort the search results to present the most relevant documents first. Assessing the performance of the ranking formula is an important part of search engine evaluation. However, the methods currently used to evaluate ranking formulae mainly collect quantitative data and do not gather qualitative data, which help to understand what needs to be improved to tailor the formulae to their end users. OBJECTIVE: This study aims to evaluate 2 different parameter settings of the ranking formula of LiSSa (the French acronym for scientific literature in health care; Department of Medical Informatics and Information), a tool that provides access to health scientific literature in French, to adapt the formula to the needs of the end users. METHODS: To collect quantitative and qualitative data, user tests were carried out with representative end users of LiSSa: 10 general practitioners and 10 registrars. Participants first assessed the relevance of the search results and then rated the ranking criteria used in the 2 formulae. Verbalizations were analyzed to characterize each criterion. RESULTS: A formula that prioritized articles representing a consensus in the field was preferred. When users assess an article's relevance, they judge its topic, methods, and value in clinical practice. CONCLUSIONS: Following the evaluation, several improvements were implemented to give more weight to articles that match the search topic and to downgrade articles that have less informative or scientific value for the reader. Applying a qualitative methodology generates valuable user inputs to improve the ranking formula and move toward a highly usable search engine.
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 | 0.339 | 0.515 |
| Meta-epidemiology (narrow) | 0.003 | 0.002 |
| Meta-epidemiology (broad) | 0.004 | 0.006 |
| Bibliometrics | 0.009 | 0.009 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.006 | 0.005 |
| Open science | 0.003 | 0.004 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 0.001 |
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".