Multi-Hop RC, HotpotQA & GNNs Select, Answer and Explain: Interpretable Multi-hop Reading Comprehension over Multiple Documents – Tu et al., AAAI 2020 Presented By: Lovish Madaan
References • HotpotQA - Peng Qi (Stanford) • GNNs - Jure Leskovec (Stanford), AAAI 2019 Tutorial by William Hamilton (McGill) • Some elements and images borrowed from Tu et al. (AAAI 2020), Yang et al. (EMNLP 2018), and Jay Alammar
Topics • Introduction and HotpotQA • Select, Answer and Explain • GNNs • Answer and Explain • Results and Ablation Study • Reviews
The Promise of Question Answering In which city was Facebook first launched? Cambridge, Massachusetts. This is because Mark Zuckerberg and his business partners launched it from his Harvard dormitory [1], and Harvard is located in Cambridge, Massachusetts [2]. [1] https://en.wikipedia.org/wiki/Mark_Zuckerberg [2] https://en.wikipedia.org/wiki/Harvard_University
Reality The Promise of Question Answering In which city was Facebook first launched? Cambridge, Massachusetts. This is because Mark Zuckerberg and his business partners launched it from his Harvard dormitory [1], and Harvard is located in Cambridge, Massachusetts [2]. [1] https://en.wikipedia.org/wiki/Mark_Zuckerberg [2] https://en.wikipedia.org/wiki/Harvard_University Sorry, folks from Google!
The Promise of Question Answering In which city was Multi-hop reasoning Facebook first launched? Cambridge, Massachusetts. This is because Mark Zuckerberg and his business partners launched it from his Harvard dormitory [1], and Harvard is located in Cambridge, Massachusetts [2]. [1] https://en.wikipedia.org/wiki/Mark_Zuckerberg [2] https://en.wikipedia.org/wiki/Harvard_University
The Promise of Question Answering In which city was Multi-hop reasoning Facebook first launched? Cambridge, Massachusetts. This is because Mark Zuckerberg and his business partners launched it from Text-based, diverse his Harvard dormitory [1], and Harvard is located in Cambridge, Massachusetts [2]. [1] https://en.wikipedia.org/wiki/Mark_Zuckerberg [2] https://en.wikipedia.org/wiki/Harvard_University
The Promise of Question Answering In which city was Multi-hop reasoning Explainability Facebook first launched? Cambridge, Massachusetts. This is because Mark Zuckerberg and his business partners launched it from Text-based, diverse his Harvard dormitory [1], and Harvard is located in Cambridge, Massachusetts [2]. [1] https://en.wikipedia.org/wiki/Mark_Zuckerberg [2] https://en.wikipedia.org/wiki/Harvard_University
Multi-hop reasoning Explainability HotpotQA Text-based, diverse Comparison Questions
Multi-hop Reasoning across Multiple Documents • Previous work (SQuAD, • HotpotQA TriviaQA, etc) When was Chris Martin born? When was the lead singer of Coldplay born? (Rajpurkar et al., 2016; Joshi et al., 2017; Dunn et al., 2017)
Explainability • Previous work • HotpotQA Answer Answer Sup fact 1 Sup fact 2
Evaluation Settings • Distractor Setting • 2 gold paragraphs + 8 extracted from information retrieval • Fullwiki Setting • Entire Wikipedia as context
• Types of Instances • Bridge Entity Questions • Comparison Questions
Topics • Introduction and HotpotQA • Select, Answer and Explain • GNNs • Answer and Explain • Results and Ablation Study • Reviews
Multi-hop RC – Previous Works • Adapt techniques from single-hop QA • Use Graph Neural Networks (GNNs) • Cao et al., 2018 – Build entity graph and realize multi-hop reasoning
Shortcomings – Previous Works • Concatenate multiple documents / Process documents separately • No document filters • Current application of GNNs • Entities as nodes – either pre specified / use NER • Further processing if answer is not an entity
Select, Answer and Explain (SAE)
Preprocessing & Inputs • Question and set of documents • Answer text • Set of labelled support sentences from each document • Label corresponding to each document - 𝐸 𝑗 (0/1) • Answer type – (“Span” / “Yes” / “No”)
Select Module • [CLS] + Q + [SEP] + D + [SEP] • One Approach – Use BCE with [CLS] embeddings as features • Neglects inter-document interactions
MHSA – Single Attention Head X – matrix of [CLS] embeddings of question/document pairs
MHSA – Multiple Attention Heads Output is the matrix of modified [CLS] embeddings having contextual information
Pairwise Bi-Linear Layer • 𝑇 𝐸 𝑗 - Score for each document (0/1/2) • 𝑚 𝑗,𝑘 = 1 𝑗𝑔 𝑇 𝐸 𝑗 > 𝑇 𝐸 𝑘 0 𝑗𝑔 𝑇 𝐸 𝑗 ≤ 𝑇(𝐸 𝑘 ) 𝑜 𝑗 • 𝑀 = − 𝑗=0 𝑘=0,𝑘≠𝑗 𝑚 𝑗,𝑘 log(𝑄 𝐸 𝑗 , 𝐸 𝑘 ) + (1 − 𝑚 𝑗,𝑘 )log(1 − 𝑄 𝐸 𝑗 , 𝐸 𝑘 ) 𝑜 Ι 𝑄 𝐸 𝑗 , 𝐸 • 𝑆 𝑗 = 𝑘 𝑘 > 0.5 - Relevance score for each document • Take top-k documents according to this relevance score
Answer Prediction • Gold Documents extracted from Select Module • [CLS] + Q + [SEP] + Context + [SEP] 𝐼 𝑗 ∈ ℝ 𝑀 × 𝑒 BERT 2-Layer 𝑍 ∈ ℝ 𝑀 × 2 MLP ( 𝑔 𝑡𝑞𝑏𝑜 )
Contextual Sentence Embeddings • Sentence Representation: • Self Attention Weights: Weighted 2-layer MLP [0/1 Label] Representation ( 𝑔 𝑏𝑢𝑢 )
Contextual Sentence Embeddings - 2 • Motivation for adding start and end span probabilities • Answer span -> Supporting Sentence • Final sentence embeddings:
Sentence Graph • Construct a graph with the following properties: • Nodes represent the sentences • Each node has label 0/1 (supporting sentence) • 3 types of edges • Between nodes present in the same document (Type 1) • Between nodes of different documents if they have named entities / noun phrases (can be different) present in the question (Type 2) • Between nodes of different documents if they have the same named entity / noun phrase (Type 3)
Sentence Graph
Topics • Introduction and HotpotQA • Select, Answer and Explain • GNNs • Answer and Explain • Results and Ablation Study • Reviews
Images T ext/Speech Modern deep learning toolbox is designed for simple sequences & grids Jure Leskovec, Stanford University 11
The Math § Average neighbor messages and apply a neural network. Initial “ layer 0 ” embeddings are previous layer equal to node features h 0 embedding of v v = x v 0 1 X h k − 1 N ( v ) | + B k h k − 1 @ W k A , 8 k > 0 h k u v = σ v | u 2 N ( v ) kth layer embedding non-linearity (e.g., average of neighbor ’ s of v ReLU or tanh) previous layer embeddings 19 Tutorial on Graph Representation Learning, AAAI 2019
Graph Attention Networks § Augment basic graph neural network model with attention. X ↵ v,u W k h k − 1 h k v = σ ( ) u u 2 N ( v ) [ { v } Non-linearity Sum over all neighbors (and the node itself) Tutorial on Graph Representation Learning, AAAI 2019 60
Training the Model § z A u Tutorial on Graph Representation Learning, AAAI 2019 20
Training the Model trainable matrices h 0 v = x v (i.e., what we learn) 0 1 X h k − 1 + B k h k − 1 @ W k A , 8 k 2 { 1 , . h k u v = σ , K } . . v | N ( v ) | u 2 N ( v ) z v = h K v § After K-layers of neighborhood aggregation, we get output embeddings for each node. § and run stochastic gradient descent to train the aggregation parameters. Tutorial on Graph Representation Learning, AAAI 2019 21
Training the Model § : Directly train the model for a supervised task (e.g., node classification): classification weights X v ✓ )) + (1 − y v ) log(1 − σ ( z > v ✓ )) y v log( σ ( z > L = v 2 V output node embedding node class label Tutorial on Graph Representation Learning, AAAI 2019 24
Overview of Model z A u Tutorial on Graph Representation Learning, AAAI 2019 25
Overview of Model Tutorial on Graph Representation Learning, AAAI 2019 26
Overview of Model Tutorial on Graph Representation Learning, AAAI 2019 27
Topics • Introduction and HotpotQA • Select, Answer and Explain • GNNs • Answer and Explain • Results and Ablation Study • Reviews
Aggregation mechanism in SAE
Graph Representation • Weighted sum of the embeddings of the nodes of the graph • The weights are given by
Answer and Explain Pipeline
Topics • Introduction and HotpotQA • Select, Answer and Explain • GNNs • Answer and Explain • Results and Ablation Study • Reviews
Dataset Details • Train – 90K • Validation/Dev – 7.4K • Test – 7.4K
Results
Ablation Study – Document Selection Module
Ablation Study – Answer & Explain Module
Ablation Study – Bridge / Comp. Questions
Attention Heatmap Example Question - “Were Scott Derrickson and Ed Wood of the same nationality?”
HotpotQA Leaderboard
Topics • Introduction and HotpotQA • Select, Answer and Explain • GNNs • Answer and Explain • Results and Ablation Study • Reviews
Recommend
More recommend