ComPer: A Comprehensive Performance Evaluation Method for Recommender Systems
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
Systems are receiving substantial attention in several application areas (such as healthcare systems and e-commerce), where each area has different requirements. These systems are multifaceted by nature. So, many metrics, which are sometimes contradictious, are introduced to assess different aspects. The existence of several alternatives and dimensions to recommendation approaches complicate the evaluation of recommender systems. In such a situation, it is desirable to evaluate and compare recommenders in a united way that assesses the multifaceted aspects of these systems fairly and uniformly. Despite the abundance of evaluation dimensions, the literature still lacks an evaluation method that evaluates the multiple properties of these systems, all at once. As a potential solution, this paper proposes an evaluation methodology that provides a multidimensional assessment of recommender systems. The proposed method, which we call ComPer, combines the most common evaluation dimensions into a single, yet, general evaluation metric. ComPer is inspired by the idea that a recommender system mimics human beings; hence, it can be seen as a human and its outputs can be assessed as human's outputs. Up to our knowledge, this is the first evaluation approach that deals with recommenders as humans. ComPer aims to be thorough (by combining multiple dimensions), simple (by presenting the final result as a single value), and independent (by providing setting-independent results). The applicability of the proposed methodology is evaluated empirically using three different datasets. The initial results are promising in the sense that ComPer is able to give comparable results regardless of the experimental settings.
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 | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.004 |
| Open science | 0.002 | 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