Question Answering by Reasoning Across Documents with Graph Convolutional Networks

Cao, Nicola De; Aziz, Wilker; Titov, Ivan

doi:10.18653/v1/n19-1240

Cited by 201 publications

(146 citation statements)

References 26 publications

Supporting

Mentioning

134

Contrasting

Order By: Relevance

“…Dhingra et al (2018) modify an existing neural QA model to additionally incorporate coreference information provided by a coreference resolution model. De Cao et al (2018) build a graph connecting entities and apply Graph Convolutional Networks (Kipf and Welling, 2016) to perform multi-hop reasoning, which leads to strong results on WIKIHOP. Zhong et al (2019) propose a new neural QA architecture that combines a combination of coarse-grained and fine-grained reasoning to achieve very strong results on WIKIHOP.…”

Section: Related Workmentioning

confidence: 99%

NLProlog: Reasoning with Weak Unification for Question Answering in Natural Language

Weber¹,

Minervini²,

Münchmeyer³

et al. 2019

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics

View full text Add to dashboard Cite

Rule-based models are attractive for various tasks because they inherently lead to interpretable and explainable decisions and can easily incorporate prior knowledge. However, such systems are difficult to apply to problems involving natural language, due to its linguistic variability. In contrast, neural models can cope very well with ambiguity by learning distributed representations of words and their composition from data, but lead to models that are difficult to interpret. In this paper, we describe a model combining neural networks with logic programming in a novel manner for solving multi-hop reasoning tasks over natural language. Specifically, we propose to use a Prolog prover which we extend to utilize a similarity function over pretrained sentence encoders. We fine-tune the representations for the similarity function via backpropagation. This leads to a system that can apply rulebased reasoning to natural language, and induce domain-specific rules from training data. We evaluate the proposed system on two different question answering tasks, showing that it outperforms two baselines -BIDAF (Seo et al., 2016a) and FASTQA (Weissenborn et al., 2017b) on a subset of the WIKIHOP corpus and achieves competitive results on the MEDHOP data set (Welbl et al., 2017).

show abstract

Section: Related Workmentioning

confidence: 99%

NLProlog: Reasoning with Weak Unification for Question Answering in Natural Language

Weber¹,

Minervini²,

Münchmeyer³

et al. 2019

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics

View full text Add to dashboard Cite

show abstract

“…In previous studies, GCNs are used to encode dependency trees (Marcheggiani and Titov, 2017;Zhang et al, 2018) and cross-document relations (Yasunaga et al, 2017;De Cao et al, 2019) for downstream tasks. Our work is the first to leverage GCNs for encoding conversation structures.…”

Section: Conversational-gcn: Aggregation-based Structure Modeling Formentioning

confidence: 99%

Modeling Conversation Structure and Temporal Dynamics for Jointly Predicting Rumor Stance and Veracity

Wei¹,

Xu²,

Mao³

2019

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen

View full text Add to dashboard Cite

Automatically verifying rumorous information has become an important and challenging task in natural language processing and social media analytics. Previous studies reveal that people's stances towards rumorous messages can provide indicative clues for identifying the veracity of rumors, and thus determining the stances of public reactions is a crucial preceding step for rumor veracity prediction. In this paper, we propose a hierarchical multi-task learning framework for jointly predicting rumor stance and veracity on Twitter, which consists of two components. The bottom component of our framework classifies the stances of tweets in a conversation discussing a rumor via modeling the structural property based on a novel graph convolutional network. The top component predicts the rumor veracity by exploiting the temporal dynamics of stance evolution. Experimental results on two benchmark datasets show that our method outperforms previous methods in both rumor stance classification and veracity prediction. Timeline 1 2 3 4 5 6Source tweet (originating a rumor): 1.Reply tweets: 2, 3, 4, 5, and 6. Support< l a t e x i t s h a 1 _ b a s e 6 4 = " L m F 7 u y l m E d U T 0 h / 2 S z Q h Z J C 2 F 1 0 = " > A A A B 7 n i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z Q O S J c x O Z p M h s z P D T K 8 Q Q j 7 C i w d F v P o 9 3 v w b J 8 k e N L G g o a j q p r s r 1 o J b D I J v r 7 C 2 v r G 5 V d w u 7 e z u 7 R + U D 4 + a V m W G s g Z V Q p l 2 T C w T X L I G c h S s r Q 0 j a S x Y K x 7 d z v z W E z O W K / m I Y 8 2 i l A w k T z g l 6 K T W Q 6 a 1 M t g r V 4 J q M I e / S s K c V C B H v V f + 6 v Y V z V I m k Q p i b S c M N E Y T Y p B T w a a l b m a Z J n R E B q z j q C Q p s 9 F k f u 7 U P 3 N K 3 0 + U c S X R n 6 u / J y Y k t X a c x q 4 z J T i 0 y 9 5 M / M / r Z J h c R x M u d Y Z M 0 s W i J B M + K n / 2 u 9 / n h l E U Y 0 c I N d z d 6 t M h M Y S i S 6 j k Q g i X X 1 4 l z Y t q G F T D + 8 t K 7 S a P o w g n c A r n E M I V 1 O A O 6 t A A C i N 4 h l d 4 8 7 T 3 4 r 1 7 H 4 v W g p f P H M M f e J 8 / n i m P v Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " L m F 7 u y l m E d U T 0 h / 2 S z Q h Z J C 2 F 1 0 = " > A A A B 7 n i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z Q O S J c x O Z p M h s z P D T K 8 Q Q j 7 C i w d F v P o 9 3 v w b J 8 k e N L G g o a j q p r s r 1 o J b D I J v r 7 C 2 v r G 5 V d w u 7 e z u 7 R + U D 4 + a V m W G s g Z V Q p l 2 T C w T X L I G c h S s r Q 0 j a S x Y K x 7 d z v z W E z O W K / m I Y 8 2 i l A w k T z g l 6 K T W Q 6 a 1 M t g r V 4 J q M I e / S s K c V C B H v V f + 6 v Y V z V I m k Q p i b S c M N E Y T Y p B T w a a l b m a Z J n R E B q z j q C Q p s 9 F k f u 7 U P 3 N K 3 0 + U c S X R n 6 u / J y Y k t X a c x q 4 z J T i 0 y 9 5 M / M / r Z J h c R x M u d Y Z M 0 s W i J B M + K n / 2 u 9 / n h l E U Y 0 c I N d z d 6 t M h M Y S i S 6 j k Q g i X X 1 4 l z Y t q G F T D + 8 t K 7 S a P o w g n c A r n E M I V 1 O A O 6 t A A C i N 4 h l d ...

show abstract

“…In both cases, all relevant information, barring some linguistic knowledge, is provided or the questions are unanswerable (Rajpurkar et al, 2018). This allows using an attention-based approach of indirectly combining information (Dhingra et al, 2018;Cao et al, 2019;Song et al, 2018).…”

Section: Related Workmentioning

confidence: 99%

What’s Missing: A Knowledge Gap Guided Approach for Multi-hop Question Answering

Khot¹,

Sabharwal

Clark³

2019

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen

View full text Add to dashboard Cite

Multi-hop textual question answering requires combining information from multiple sentences. We focus on a natural setting where, unlike typical reading comprehension, only partial information is provided with each question. The model must retrieve and use additional knowledge to correctly answer the question. To tackle this challenge, we develop a novel approach that explicitly identifies the knowledge gap between a key span in the provided knowledge and the answer choices. The model, GapQA, learns to fill this gap by determining the relationship between the span and an answer choice, based on retrieved knowledge targeting this gap. We propose jointly training a model to simultaneously fill this knowledge gap and compose it with the provided partial knowledge. On the Open-BookQA dataset, given partial knowledge, explicitly identifying what's missing substantially outperforms previous approaches.

show abstract

Question Answering by Reasoning Across Documents with Graph Convolutional Networks

Cited by 201 publications

References 26 publications

NLProlog: Reasoning with Weak Unification for Question Answering in Natural Language

NLProlog: Reasoning with Weak Unification for Question Answering in Natural Language

Modeling Conversation Structure and Temporal Dynamics for Jointly Predicting Rumor Stance and Veracity

What’s Missing: A Knowledge Gap Guided Approach for Multi-hop Question Answering

Contact Info

Product

Resources

About