Browsing by Author "Gong, Jiaying"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Few-Shot and Zero-Shot Learning for Information Extraction
    Gong, Jiaying (Virginia Tech, 2024-05-31)
    Information extraction aims to automatically extract structured information from unstructured texts. Supervised information extraction requires large quantities of labeled training data, which is time-consuming and labor-intensive. This dissertation focuses on information extraction, especially relation extraction and attribute-value extraction in e-commerce, with few labeled (few-shot learning) or even no labeled (zero-shot learning) training data. We explore multi-source auxiliary information and novel learning techniques to integrate semantic auxiliary information with the input text to improve few-shot learning and zero-shot learning. For zero-shot and few-shot relation extraction, the first method explores the existing data statistics and leverages auxiliary information including labels, synonyms of labels, keywords, and hypernyms of name entities to enable zero-shot learning for the unlabeled data. We build an automatic hypernym extraction framework to help acquire hypernyms of different entities directly from the web. The second method explores the relations between seen classes and new classes. We propose a prompt-based model with semantic knowledge augmentation to recognize new relation triplets under the zero-shot setting. In this method, we transform the problem of zero-shot learning into supervised learning with the generated augmented data for new relations. We design the prompts for training using the auxiliary information based on an external knowledge graph to integrate semantic knowledge learned from seen relations. The third work utilizes auxiliary information from images to enhance few-shot learning. We propose a multi-modal few-shot relation extraction model that leverages both textual and visual semantic information to learn a multi-modal representation jointly. To supplement the missing contexts in text, this work integrates both local features (object-level) and global features (pixel-level) from different modalities through image-guided attention, object-guided attention, and hybrid feature attention to solve the problem of sparsity and noise. We then explore the few-shot and zero-shot aspect (attribute-value) extraction in the e-commerce application field. The first work studies the multi-label few-shot learning by leveraging the auxiliary information of anchor (label) and category description based on the prototypical networks, where the hybrid attention helps alleviate ambiguity and capture more informative semantics by calculating both the label-relevant and query-related weights. A dynamic threshold is learned by integrating the semantic information from support and query sets to achieve multi-label inference. The second work explores multi-label zero-shot learning via semi-inductive link prediction of the heterogeneous hypergraph. The heterogeneous hypergraph is built with higher-order relations (generated by the auxiliary information of user behavior data and product inventory data) to capture the complex and interconnected relations between users and the products.
  • Thumbnail Image
    Knowledge-Enhanced Multi-Label Few-Shot Product Attribute-Value Extraction
    Gong, Jiaying; Chen, Wei-Te; Eldardiry, Hoda (ACM, 2023-10-21)
    Existing attribute-value extraction (AVE) models require large quantities of labeled data for training. However, new products with new attribute-value pairs enter the market every day in real-world e- Commerce. Thus, we formulate AVE in multi-label few-shot learning (FSL), aiming to extract unseen attribute value pairs based on a small number of training examples. We propose a Knowledge- Enhanced Attentive Framework (KEAF) based on prototypical networks, leveraging the generated label description and category information to learn more discriminative prototypes. Besides, KEAF integrates with hybrid attention to reduce noise and capture more informative semantics for each class by calculating the label-relevant and query-related weights. To achieve multi-label inference, KEAF further learns a dynamic threshold by integrating the semantic information from both the support set and the query set. Extensive experiments with ablation studies conducted on two datasets demonstrate that our proposed model significantly outperforms other SOTA models for information extraction in few-shot learning.
  • Thumbnail Image
    Multi-Label Zero-Shot Product Attribute-Value Extraction
    Gong, Jiaying; Eldardiry, Hoda (ACM, 2024-05-13)
    E-commerce platforms should provide detailed product descriptions (attribute values) for effective product search and recommendation. However, attribute value information is typically not available for new products. To predict unseen attribute values, large quantities of labeled training data are needed to train a traditional supervised learning model. Typically, it is difficult, time-consuming, and costly to manually label large quantities of new product profiles. In this paper, we propose a novel method to efficiently and effectively extract unseen attribute values from new products in the absence of labeled data (zero-shot setting).We propose HyperPAVE, a multilabel zero-shot attribute value extraction model that leverages inductive inference in heterogeneous hypergraphs. In particular, our proposed technique constructs heterogeneous hypergraphs to capture complex higher-order relations (i.e. user behavior information) to learn more accurate feature representations for graph nodes. Furthermore, our proposed HyperPAVE model uses an inductive link prediction mechanism to infer future connections between unseen nodes. This enables HyperPAVE to identify new attribute values without the need for labeled training data. We conduct extensive experiments with ablation studies on different categories of the MAVE dataset. The results demonstrate that our proposed HyperPAVE model significantly outperforms existing classificationbased, generation-based large language models for attribute value extraction in the zero-shot setting.
  • Thumbnail Image
    Text Analytics and Machine Learning (TML) CS5604 Fall 2019
    Mansur, Rifat Sabbir; Mandke, Prathamesh; Gong, Jiaying; Bharadwaj, Sandhya M.; Juvekar, Adheesh Sunil; Chougule, Sharvari (Virginia Tech, 2019-12-29)
    In order to use the burgeoning amount of data for knowledge discovery, it is becoming increasingly important to build efficient and intelligent information retrieval systems.The challenge in informational retrieval lies not only in fetching the documents relevant to a query but also in ranking them in the order of relevance. The large size of the corpora as well as the variety in the content and the format of information pose additional challenges in the retrieval process. This calls for the use of text analytics and machine learning techniques to analyze and extract insights from the data to build an efficient retrieval system that enhances the overall user experience. With this background, the goal of the Text Analytics and Machine Learning team is to suitably augment the document indexing and demonstrate a qualitative improvement in the document retrieval. Further, we also plan to make use of document browsing and viewing logs to provide meaningful recommendations to the user. The goal of the class is to build an end-to-end information retrieval system for two document corpora, viz., Electronic Theses & Dissertations (ETDs) and Tobacco Settlement Records (TSRs). The ETDs are a collection of over 33,000 thesis and dissertation documents in VTechWorks at Virginia Tech. The challenge in building a retrieval system around this corpus lies in the distinct nature of ETDs as opposed to other well studied document formats such as conference/journal publications and web-pages. The TSR corpus consists of over 14M records covering formats ranging from letters and memos to image based advertisements. We seek to understand the nature of both these corpora as well as the information need patterns of the users in order to augment the index based search with domain specific information using machine learning based methods. Extending prior experiments, we investigate reasons for the unbalanced nature of the clusters from the previous iterations of the K-Means algorithm on the tobacco data. In addition, we explore and present preliminary results of running Agglomerative Clustering on a small subset of the tobacco data. We also explored different pre-trained models of detecting sentiments. We identified a package, empath, that shows better results in identifying emotions in the tobacco deposition documents. Besides, we implemented text summarization based on both Latent Semantic Analysis and the Luhn Algorithm on the tobacco (article) data (38,038 documents). We also implemented text summarization on a sample ETD chapter dataset.