Stacking Outperforms in Debiased Neural Collaborative Filtering: A Comparative Study of IPS-Weighted NCF and Tree-Based Models for Exposure-Biased CTR Prediction
Bibliographic record
Abstract
Recent developments in recommender systems have increasingly employed deep learning methodologies to confront long-standing challenges, including the modeling of intricate user–item interactions, the incorporation of temporal dynamics, and the mitigation of exposure bias. This study reviews and extends insights from four representative approaches. First, the Convolutional Transformer Neural Collaborative Filtering (CTNCF) model combines convolutional neural networks with Transformer architectures to capture both localized and long-range dependencies within user–item representations, thereby surpassing the performance of conventional Neural Collaborative Filtering (NCF). Second, the Neural Tensor Factorization (NTF) framework advances classical tensor factorization by embedding recurrent and multilayer neural components, enabling the representation of time-varying preferences and nonlinear interactions among latent factors. Third, the Deep Interest Network (DIN) introduces a local activation mechanism that adaptively models user interests in click-through rate prediction, effectively overcoming the limitations of fixed-length embeddings in capturing heterogeneous behavioral patterns; notably, this model has been deployed at scale in industrial advertising contexts. Finally, recent work addressing de-exposure bias in NCF incorporates reward signals derived from the LinUCB algorithm into the neural recommendation process, thereby enhancing both fairness and predictive accuracy by increasing the visibility of underexposed items. Taken together, these contributions illustrate the progression of neural recommender systems from static factorization paradigms toward dynamic, adaptive, and fairness-oriented frameworks, offering both theoretical contributions and practical value for the design of large-scale recommendation platforms.
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 | 0.021 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.004 |
| Open science | 0.003 | 0.001 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 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".