Personalizing E-Commerce by Optimizing LLMs for Tailored Product Recommendations
Bibliographic record
Abstract
With rapid advancements in Large Language Model (LLM) technology and the emergence of open-source implementations for running these models, researchers have increasingly turned to LLMs to enhance the capabilities of traditional Machine Learning methods, especially within the purview of product recommendation system. In a paper by Sun et al, they implemented a novel framework which used an LLM integrated with collaborative filtering to determine product recommendations [1]. But this approach relies on a decoder to generate final product recommendations and ranking. Other researchers focused on comparing Encoder and Decoder architectures showed that a fine-tuned encoder can significantly outperform decoder architectures in language understanding tasks [2]. Thus, we propose a novel framework for product recommendation which incorporates both the real world context knowledge and reasoning capabilities of decoder architectures coupled with the superior language capabilities of encoder architectures. Siamese Dual Encoders (SDA) have long been used for tasks including Question-Answering, sentiment analysis, and generating recommendations [3]. By preprocessing product review data using a reasoning based LLM, and then using the processed data to train a BERT-based Siamese Dual Encoder, we aim to demonstrate this novel method which yields better results in terms of more accurate recommendations. This paper will show the positive impact of data preprocessing using LLMs to incorporate real world context and reasoning capabilities, while also integrating the ranking and classification capabilities of a BERT-based encoder.
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 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 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 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
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".