Improving Relevance in a Recommendation System to Suggest Charities without Explicit User Profiles Using Dual-Autoencoders
Bibliographic record
Abstract
This work explores the effect of the quality of inferred user profiles on the accuracy of charitable recommendations when using an item-based collaborative filter algorithm. A gap was identified in the literature with respect to the application of charitable recom-mendation systems in the absence of rich user profiles. This paper introduces an approach to generate relevant recommendations when neither user profiles nor feedback on donation preferences is available. The discovery of user preferences is achieved via the construction of implicit ratings computed from custom feature engineering, while the sparsity of item and user ratings was addressed with a dimension reduction strategy based on dual-autoencoders from a commercial machine learning platform. Our analysis shows the magnitude and sensitivity of the relationship between the relevance of the recommendations and the average number of donations per user. Raw data for this research was provided by a leading online donation platform and contains 24 million anonymous donations to 165 thousand unique causes from over 1.2 million users. We find that the most effective way to increase the relevance of recommendations by a factor of 2 at any top- k value is to train the collaborative filter with users that have at least 50 donations in the data set. As a result, the training set for the collaborative filter is restricted to 8% of the original users, 70% of the companies, 49% of the causes, and 70% of the original countries where users making donations reside.
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 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 | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".