MétaCan
Menu
Back to cohort
Record W4413175548 · doi:10.18280/ts.420438

Cross-Modal Product Image Retrieval for E-Commerce Recommendation Systems via Deep Learning

2025· article· en· W4413175548 on OpenAlexvenueno aff
Chao Zhang

Bibliographic record

VenueTraitement du signal · 2025
Typearticle
Languageen
FieldComputer Science
TopicImage Retrieval and Classification Techniques
Canadian institutionsnot available
FundersGovernment of Jiangsu Province
KeywordsModalComputer scienceProduct (mathematics)Deep learningImage (mathematics)Artificial intelligenceInformation retrievalComputer visionMathematicsMaterials science

Abstract

fetched live from OpenAlex

With the rapid evolution of e-commerce, increasing demands have been placed on the accuracy of product recommendation systems.Cross-modal product image retrieval, which serves as a critical bridge between visual content and textual descriptions, has a direct impact on user experience in such systems.However, existing retrieval approaches have been predominantly tailored for natural image scenarios, rendering them less effective in ecommerce contexts.Product images are required to highlight "purchasability features" but are often compromised by complex backgrounds, while textual descriptions-rich in both marketing language and functional attributes-frequently suffer from semantic dilution due to redundant content.Furthermore, generic models fail to optimize for the implicit alignment between "product attributes" and "user intent," resulting in suboptimal recommendation relevance.Although self-attention mechanisms have been introduced in cross-modal tasks, limitations persist in local visual feature extraction, core semantic focus in text, and deep alignment across modalities.To address these challenges, a cross-modal product image retrieval method tailored for e-commerce recommendation systems was proposed.A fullchain model based on self-attention mechanisms was constructed.A global-local visual feature extraction module was designed to enhance discriminative product regions using category labels.A textual feature extraction module was incorporated to suppress nonessential information and emphasize semantically decisive elements.Furthermore, a visionlanguage cross-extraction module was constructed to establish bidirectional mappings between sub-image patches and key textual tokens.This approach enables precise multimodal alignment, thereby providing a three-tiered matching rationale-image detail, textual demand, and purchase intent-for e-commerce recommendation systems.

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 imitation

Not 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.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Bench or experimental · Consensus signal: none
GenreCandidate signal: Methods · Consensus signal: none
Teacher disagreement score0.957
Threshold uncertainty score0.762

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.001
Science and technology studies0.0000.000
Scholarly communication0.0010.001
Open science0.0010.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.017
GPT teacher head0.295
Teacher spread0.278 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one teacher head, not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designBench or experimental
Domainnot available
GenreMethods

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".

Quick stats

Citations0
Published2025
Admission routes1
Has abstractyes

Explore more

Same venueTraitement du signalSame topicImage Retrieval and Classification TechniquesFrench-language works237,207