Learning Recommender Systems with Deep Structured Low Rank Matrix Approximation
Bibliographic record
Abstract
Collaborative filtering (CF) is a successful approach in developing recommender systems for many real-world problems.Traditional CF-based methods exploit user-item matrix for learning latent factors to make recommendations.Despite their success, data sparsity and cold start problems limit the efficiency of CF-based methods in learning effective latent factors.A classical approach to address these problems is by taking advantage of user (item) auxiliary information.Due to the natural ability of deep learning in extracting features from multiple sources of data simultaneously, it is the most common technique for integrating rating data and auxiliary information.In spite of the success of deep learning in extracting complex and non-linear features from multiple sources of data, sparsity of auxiliary information degrades the quality of the learned latent factors.Another approach that recently gained much attention in improving the performance of CF-based methods assumes the rating matrix is composed of several local regions within each of which users have similar preferences.In this technique, the rating matrix decomposes into several sub-matrices and is approximated locally.This technique however relies on the rating matrix as the sole source of information for learning, and the sparsity of the rating matrix reduces the effectiveness of the learned latent factors.To address the above problems, we propose a Deep structured LOw Rank Matrix Approximation model (DLORMA) that incorporates additional stacked denoising autoencoders and local matrix approximations in a loosely coupled fashion.To the best of our knowledge, DLORMA is the first hybrid recommendation system that combines deep learning and low rank matrix approximation.Experiments conducted on three real datasets show improvements in prediction performance over the component approaches individually.
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 | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".