An NLP-Deep Learning Approach for Product Rating Prediction Based on Online Reviews and Product Features
Bibliographic record
Abstract
This study focuses on predicting the popularity of a product based on its overall rating score. Unlike previous studies that focus on predicting the review rating based on sentiment analysis and the polarity of the reviews, in this research, the effect of product features in determining its popularity is directly measured and analyzed. To this end, a methodology consisting of three phases is considered. Phase 1 predicts the overall rating by feeding the general product features, extracted from the online product information available on Amazon webpages to three different deep learning (DL) models: deep feedforward neural network (DFFNN), probabilistic neural network (PNN), and radial basis function neural network (RBFNN). Phase 2 identifies other features that customers care about the most by applying the named entity recognition (NER) algorithm to the customer online reviews. Finally, Phase 3 feeds the combination of the general and custom features to the same DL models to predict the overall rating score of the product. The experimental results on a dataset of laptop products indicate an impressive performance of the proposed approach, which is mainly attributed to including custom product features in the inputs of the DL algorithm. More precisely, the proposed model could achieve the highest accuracy score of 84.01%, 84.68% for recall, 87.63% for precision, and 84.06% for F1 score. Applying this procedure could help businesses identify the specific areas of strengths and weaknesses of their products or services from the perspective of their customers, allowing them to thrive in today’s competitive markets.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| 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".