Improving e-commerce product recommendation using semantic context and sequential historical purchases
Bibliographic record
Abstract
Collaborative Filtering (CF)-based recommendation methods suffer from (i) sparsity (have low user–item interactions) and (ii) cold start (an item cannot be recommended if no ratings exist). Systems using clustering and pattern mining (frequent and sequential) with similarity measures between clicks and purchases for next-item recommendation cannot perform well when the matrix is sparse, due to rapid increase in number of items. Additionally, they suffer from: (i) lack of personalization: patterns are not targeted for a specific customer and (ii) lack of semantics among recommended items: they can only recommend items that exist as a result of a matching rule generated from frequent sequential purchase pattern(s). To better understand users’ preferences and to infer the inherent meaning of items, this paper proposes a method to explore semantic associations between items obtained by utilizing item (products’) metadata such as title, description and brand based on their semantic context (co-purchased and co-reviewed products). The semantics of these interactions will be obtained through distributional hypothesis, which learns an item’s representation by analyzing the context (neighborhood) in which it is used. The idea is that items co-occurring in a context are likely to be semantically similar to each other (e.g., items in a user purchase sequence). The semantics are then integrated into different phases of recommendation process such as (i) preprocessing, to learn associations between items, (ii) candidate generation, while mining sequential patterns and in collaborative filtering to select top-N neighbors and (iii) output (recommendation). Experiments performed on publically available E-commerce data set show that the proposed model performed well and reflected user preferences by recommending semantically similar and sequential products.
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 | 0.000 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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".