Question Answering with Subgraph Embeddings

By Antoine Bordes, Sumit Chopra, Jason Aaron Edward Weston. 2014

736 citations recorded by the source corpus.

This paper presents a system which learns to answer questions on a broad range of topics from a knowledge base using few hand-crafted features. Our model learns low-dimensional embeddings of words and knowledge base constituents; these representations are used to score natural language questions against candidate answers. Training our system using pairs of questions and structured representations of their answers, and pairs of question paraphrases, yields competitive results on a recent benchmark of the literature.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Natural Language Processing

Citations in the corpus

Key-Value Memory Networks for Directly Reading Documents (2016, methodology): “We compare four approaches: (i) the QA system of Bordes et al. (2014) that performs well on existing datasets We-bQuestions (Berant et al. , 2013) and SimpleQues-tions (Bordes et al. , 2015) that use KBs only; (ii) supervised embeddings that do not make use of a KB at all but learn question to…”
OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge (2019, background): “Knowledge-based question answering has received much more attention in the NLP community (e.g., [5, 50, 49, 6, 40, 28, 7]).”
Stacked Attention Networks for Image Question Answering (2015, background): “Unlike pure language based QA systems that have been studied extensively in the NLP community [28, 14, 4, 31, 3, 32], image QA systems are designed to automatically answer natural language questions according to the content of a reference image.”
Constructing Datasets for Multi-hop Reading Comprehension Across Documents (2017, methodology): “oducing synthetic links via dense latent embeddings. Several other multi-fact inference methods based on dense representations have been proposed, using composition functions such as vector addition (Bordes et al., 2014), RNNs (Neelakantan et al., 2015; Das et al., 2016), and memory networks (Jain, ”
Large-scale Simple Question Answering with Memory Networks (2015, methodology): “As in (Bordes et al., 2014a; Bordes et al., 2014b), learning is performed with gradient descent, so that negative examples (non-supporting facts or non-paraphrases) are generated according to a randomized policy during training.”
Towards Universal Paraphrastic Sentence Embeddings (2015, background): “…input representations in order to maximize similarity in the learned space (Weston et al., 2010; Yih et al., 2011; Huang et al., 2013; Hermann & Blunsom, 2014; Socher et al., 2014; Faruqui & Dyer, 2014; Bordes et al., 2014b;a; Lu et al., 2015), including our prior work (Wieting et al., 2015).”
Learning to Compose Neural Networks for Question Answering (2016, background): “These include approaches that model the task directly as a multiclass classiﬁ-cation problem (Iyyer et al., 2014), models that attempt to embed questions and answers in a shared vector space (Bordes et al., 2014) and attentional models that select words from documents sources (Hermann et al., 2015).”
Ask, Attend and Answer: Exploring Question-Guided Spatial Attention for Visual Question Answering (2015, methodology): “Existing methods include extracting key words from the question and answering based on the result returned by a search engine [29]; parsing the question and using it as a knowledge base (KB) query [2]; embedding the question and using similarity measurement to find evidence for the answer [3].”
Neural Programmer: Inducing Latent Programs with Gradient Descent (2015, methodology): “Neural networks have also been used for Q&A on datasets that do not require complicated arithmetic and logic reasoning (Bordes et al., 2014; Iyyer et al., 2014; Sukhbaatar et al., 2015; Peng et al., 2015; Hermann et al., 2015).”
Visual Madlibs: Fill in the blank Image Generation and Question Answering (2015, background): “Recently, embedding and deep learning methods have shown great promise for question-answering [30, 3, 4].”
Evaluating Prerequisite Qualities for Learning End-to-End Dialog Systems (2015, methodology): “We chose the method of Bordes et al. (2014) 10 as our baseline.”
Neural Variational Inference for Text Processing (2015, background): “One of the popular streams is mapping factoid questions with answer triples in the knowledge base (Bordes et al., 2014a;b; Yih et al., 2014).”
A Comprehensive Survey of Graph Embedding: Problems, Techniques, and Applications (2017, background): “[94] utilize a knowledge base to enrich the information about the answer.”
Video Paragraph Captioning Using Hierarchical Recurrent Neural Networks (2015, background): “The average strategy is inspired by the QA embedding [5] in which questions and answers are both represented as a combination of the embeddings of their individual words and/or symbols.”
Deep Learning for Answer Sentence Selection (2014, methodology): “[3, 2] used a type of siamese network for le arning to project question and answer pairs into a joint space.”
Machine Comprehension with Syntax, Frames, and Semantics (2015, background): “Recent question answering (QA) systems (Ferrucci et al., 2010; Berant et al., 2013; Bordes et al., 2014) have focused on open-domain factoid”
Explicit Knowledge-based Reasoning for Visual Question Answering (2015, methodology): “ailable. The advances in structured KBs, have driven an increasing interest in the NLP and AI communities in the problem of natural language question answering using structured KBs (refer to as KBQA) [4, 7, 8, 16, 23, 25, 41, 26, 38]. The VQA approach which is closest to KB-QA (and to our approach) ”
Semantic Parsing via Staged Query Graph Generation: Question Answering with Knowledge Base (2015, methodology): “Our CNN models can be analogous to the embedding approaches (Bordes et al., 2014a; Yang et al., 2014), but are more sophisticated.”
Trick Me If You Can: Human-in-the-Loop Generation of Adversarial Examples for Question Answering (2018, methodology): “ﬁcult for models. Though Wikipedia and Quizbowl are informationrich, they do not contain all the information necessary to answer every question. QA systems that integrate a knowledge base end-to-end (Bordes et al., 2014) or use web queries (Chen et al., 2017) could better answer questions with unsee”