Résultats de recherche pour « Vector embeddings »

Stock embeddings learned from correlation networks or return series are increasingly used as compact descriptors of asset similarity. We ask a simple question: do these stock embeddings outperform or tie with the raw features they compress? We benchmark a modular pipeline (relationship matrix, graph construction, embedding generation, evaluation) across three universes (S&P 500, Russell 1000, Russell 3000), four horizons, six co-movement-removal methods, five graph configurations, and six representation families-including a faithful reproduction of Sarmah et. al.-evaluating retrieval (hit@k, success@k, MRR), clustering (V-measure, NMI, ARI), and supervised readout (SVM-F1) against GICS labels, with paired Wilcoxon and McNemar tests, bootstrap intervals, and Benjamini-Hochberg correction. The answer is baseline-dependent. Every learned family decisively beats a leak-free raw-correlation-row baseline (+0.11 to +0.21 in sub-industry hit@3, p ≪ 0.001), but none clearly beats the raw return series: at Russell 3000 the best embedding is statistically tied with raw returns (p = 0.69), and elsewhere raw returns leads by at most 0.027-0.029 in hit@3. Under a decision rule requiring BH-significance plus one percentage point, raw returns never loses. Embeddings' durable advantage is compression: 16-64 dimensions match thousands of raw features. An out-of-sample peer-hedging experiment shows no risk-reduction advantage over a direct Pearson hedge. All results use full-survivor panels and compare methods under idealised conditions, not live trading.

Platform kursus digital masih menghadapi tingkat penyelesaian belajar yang rendah karena pengguna sering kesulitan menentukan urutan materi yang sesuai dengan tujuan, minat, dan kemampuan awal. Personalisasi jalur belajar menjadi pendekatan yang relevan, namun menghadapi kendala cold-start problem ketika riwayat interaksi pengguna belum tersedia. Penelitian ini mengimplementasikan fitur rekomendasi jalur belajar berbasis semantic search menggunakan model text-embedding-3-small dari OpenAI pada platform kursus digital berbasis web bernama Genii. Setiap kursus direpresentasikan dari judul, deskripsi, topik, tingkat kesulitan, dan ringkasan lesson, lalu dikonversi menjadi embeddingberdimensi 1536 dan diindeks pada layanan vector search Convex. Rekomendasi dihasilkan dengan membentuk embedding preferensi pengguna, menghitung kemiripan menggunakan cosine similarity, dan mengambil Top-K kursus dengan skor tertinggi. Evaluasi penerimaan pengguna dilakukan menggunakan Technology Acceptance Model (TAM) terhadap 50 responden dari 14 provinsi. Hasil menunjukkan nilai rata-rata Perceived Usefulness (PU) 4,40, Perceived Ease of Use (PEOU) 4,33, dan Behavioral Intention (BI) 4,38 pada skala 1–5, dengan reliabilitas konstruk Cronbach’s alpha di atas 0,70. Korelasi Pearson menunjukkan hubungan positif signifikan antara PU dan BI (r = 0,7324, p < 0,001) serta PEOU dan BI (r = 0,6847, p < 0,001). Penelitian ini berkontribusi pada pembuktian bahwa pendekatan semantic search berbasis embedding dapat diimplementasikan secara realistis pada platform pendidikan digital dan dipersepsikan bermanfaat serta mudah digunakan oleh pengguna.

Bitcoin Blockchain is a completely public open currency transaction ledger, recent growth in the crypto-currency market has driven many lawful and blackmarket actors using Bitcoin as the main method of payment. I propose address2vec an algorithm to generate vector representations of addresses on the Bitcoin Blockchain.I am able to present better results than a baseline random approach in predicting hoarding vs spending beaviour. The current work allows for utilization of common machine learning algorithms on bitcoin transaction addresses.

Embedding social network data into a low-dimensional vector space has shown promising performance for many real-world applications, such as node classification, node clustering, link prediction and network visualization. However, the information contained in these vector embeddings remains abstract and hard to interpret. Methods for inspecting embeddings usually rely on visualization methods, which do not work on a larger scale and do not give concrete interpretations of vector embeddings in terms of preserved network properties (e.g., centrality or betweenness measures). In this paper, we study and investigate network properties preserved by recent random walk-based embedding procedures like node2vec, DeepWalk or LINE. We propose a method that applies learning to rank in order to relate embeddings to network centralities. We evaluate our approach with extensive experiments on real-world and artificial social networks. Experiments show that each embedding method learns different network properties. In addition, we show that our graph embeddings in combination with neural networks provide a computationally efficient way to approximate the Closeness Centrality measure in social networks.

<div>We develop and evaluate an automated data-driven framework for providing reviewer recommendations for submitted manuscripts. Given inputs comprising a set of manuscripts for review and a listing of a pool of prospective reviewers, our system uses a publisher database to extract papers authored by the reviewers from which a Paragraph Vector (doc2vec ) neural network model is learned and used to obtain vector space embeddings of documents. Similarities between embeddings of an individual reviewer’s papers and a manuscript are then used to compute manuscript-reviewer match scores and to generate a ranked list of recommended reviewers for each manuscript. Our mainline proposed system uses full text versions of the reviewers’ papers, which we demonstrate performs significantly better than models developed based on abstracts alone, which has been the predominant paradigm in prior work. Direct retrieval of reviewer’s manuscripts from a publisher database reduces reviewer burden, ensures up-to-date data, and eliminates the potential for misuse through data manipulation. We also propose a useful evaluation methodology that addresses hyperparameter selection and enables indirect comparisons with alternative approaches and on prior datasets. Finally, the work also contributes a large scale retrospective reviewer matching dataset and evaluation that we hope will be useful for further research in this field. Our system is quite effective; for the mainline approach, expert judges rated 38% of the recommendations as Very Relevant; 33% as Relevant; 24% as Slightly Relevant; and only 5% as Irrelevant.</div>

This work addresses the need for efficient and reusable vector embeddings (VEs) for large-scale RDF Knowledge Graphs, focusing on the widely used DrugBank dataset. The study aims to reduce the computational burden and environmental impact associated with repeatedly generating embeddings for downstream tasks such as link prediction, clustering, and recommendation. To this end, embeddings are generated from the DrugBank Knowledge Graph—comprising over 3.6 million triples, more than 1.5 million entities, and 95 relations—using 25 models implemented in the PyKEEN framework. The dataset is processed into training, validation, and test splits, and embeddings of fixed dimensionality are produced for both entities and relations. The resulting representations are released in multiple formats, including full JSON files, class-partitioned subsets, and an efficient Parquet-based structure, to support scalable querying. Experimental evaluation demonstrates substantial improvements in access time and memory usage when using structured formats, particularly Parquet. Additionally, the study quantifies the carbon footprint of embedding generation, showing that distributing precomputed embeddings can reduce energy consumption by up to 99.37% compared to on-demand recomputation. Overall, the work provides a comprehensive, reusable resource that facilitates research while promoting computational efficiency and environmental sustainability.

We develop and evaluate an automated data-driven framework for providing reviewer recommendations for submitted manuscripts. Given inputs comprising a set of manuscripts for review and a listing of a pool of prospective reviewers, our system uses a publisher database to extract papers authored by the reviewers from which a Paragraph Vector (doc2vec ) neural network model is learned and used to obtain vector space embeddings of documents. Similarities between embeddings of an individual reviewer’s papers and a manuscript are then used to compute manuscript-reviewer match scores and to generate a ranked list of recommended reviewers for each manuscript. Our mainline proposed system uses full text versions of the reviewers’ papers, which we demonstrate performs significantly better than models developed based on abstracts alone, which has been the predominant paradigm in prior work. Direct retrieval of reviewer’s manuscripts from a publisher database reduces reviewer burden, ensures up-to-date data, and eliminates the potential for misuse through data manipulation. We also propose a useful evaluation methodology that addresses hyperparameter selection and enables indirect comparisons with alternative approaches and on prior datasets. Finally, the work also contributes a large scale retrospective reviewer matching dataset and evaluation that we hope will be useful for further research in this field. Our system is quite effective; for the mainline approach, expert judges rated 38% of the recommendations as Very Relevant; 33% as Relevant; 24% as Slightly Relevant; and only 5% as Irrelevant.

Compound similarity is fundamental to various cheminformatics analyses, particularly in the drug discovery industry, where the structure-activity principle is central to medicinal chemistry. Historically, binary fingerprints combined with Tanimoto and “Tanimoto-related metrics” (such as Dice, Sørensen–Dice, and Tversky) have been the gold standard for compound similarity measures. However, with the rise of large language models (LLMs) and retrieval- augmented generation (RAG), the use of embeddings for molecular representation, storage, and search has gained attention. In this study, we systematically compare molecular embeddings generated by different models, including autoencoders (AE), graph convolutional neural networks (GCNN), BERT-like models, word2vec-like and molecular attention transformers (MAT). We evaluate the performance of these embeddings against the widely used ECFP fingerprints in terms of similarity searching and clustering. To facilitate efficient searches, we leverage vector databases, which have been well established in the fields of natural language processing and RAG. Our findings demonstrate that Continuous Data-Driven Descriptors (CDDD) and MolFormer outperform tradi- tional methods in terms of similarity search efficiency and speed. This research contributes to advancing molecular representation techniques in drug discovery, potentially accelerating the identification of promising drug candidates.

ABSTRACT Background Interpreting the pathogenicity of genetic variants remains a critical bottleneck in genomic medicine. Millions of variants of uncertain significance (VUS) hinder the clinical application of genetic findings. Traditional computational approaches often rely on hand-engineered features and fail to fully capture the complexity of multidimensional genomic annotations. Methods We developed a novel semantic embedding framework, VUS.Life, that transforms variant annotations into natural language descriptions and leverages pre-trained language models to capture pathogenicity-relevant relationships in high-dimensional vector space. This approach enables direct pathogenicity prediction through representation learning rather than traditional feature-based methods. Results We evaluated the framework using curated variants from three disease-associated genes: BRCA1 (n=3,311) and BRCA2 (n=4,074) from BRCA Exchange, and FBN1 (n=1,532) from ClinVar. Variant annotations from the Variant Effect Predictor (VEP) were converted into natural language while excluding fields linked to known pathogenicity assertions to prevent data leakage. We then embedded these descriptions using three models: MPNet (all-mpnet-base-v2), Google’s text-embedding-004, and MedEmbed-large-v0.1. A k-nearest neighbor (k-NN) approach (up to 20 neighbors) was used to predict pathogenicity for new or unreviewed variants. Dimensionality reduction techniques (PCA, t-SNE, UMAP) enabled visualization of the embedding spaces. k-NN classification showed exceptional performance across all genes and embedding models. For BRCA1 , overall accuracy ranged from 97.3% (Google) to 97.9% (MPNet), with benign/likely benign accuracy of 95.1–97.2% and pathogenic/likely pathogenic accuracy of 97.9–98.4%. For BRCA2 , overall accuracy ranged from 97.9% (Google) to 99.1% (MedEmbed), with benign/likely benign accuracy of 96.6–99.2% and pathogenic/likely pathogenic accuracy of 98.6–99.4%. FBN1 validation confirmed the method’s generalizability, with accuracy exceeding 96% across all embeddings. Application to not-yet-reviewed BRCA1/2 variants demonstrated the framework’s practicality and scalability, with unknown variants aligning closely to known benign or pathogenic clusters. Conclusions This semantic embedding framework, VUS.Life, accurately captures pathogenicity-relevant features from complex variant annotations, enabling high-accuracy (>96%) automated classification across multiple genes and models. The approach generalizes beyond well-curated genes and supports scalable, interpretable, and representation-based classification of VUS. It holds significant promise for alleviating the variant interpretation bottleneck in clinical genomics.