Filters








328 Hits in 4.7 sec

Using Local Knowledge Graph Construction to Scale Seq2Seq Models to Multi-Document Inputs [article]

Angela Fan, Claire Gardent, Chloe Braud, Antoine Bordes
2019 arXiv   pre-print
We propose constructing a local graph structured knowledge base for each query, which compresses the web search information and reduces redundancy.  ...  For two generative tasks with very long text input, long-form question answering and multi-document summarization, feeding graph representations as input can achieve better performance than using retrieved  ...  Rather than encoding a static knowledge graph or leveraging external knowledge graphs, we build a local graph for each query and model these using standard Seq2Seq models.  ... 
arXiv:1910.08435v1 fatcat:2epldhegyfbsxmqtb5jngtfgra

Using Local Knowledge Graph Construction to Scale Seq2Seq Models to Multi-Document Inputs

Angela Fan, Claire Gardent, Chloé Braud, Antoine Bordes
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
We propose constructing a local graph structured knowledge base for each query, which compresses the web search information and reduces redundancy.  ...  Current approaches extractively select portions of web text as input to Sequence-to-Sequence models using methods such as TF-IDF ranking.  ...  Rather than encoding a static knowledge graph or leveraging external knowledge graphs, we build a local graph for each query and model these using standard Seq2Seq models.  ... 
doi:10.18653/v1/d19-1428 dblp:conf/emnlp/FanGBB19 fatcat:u4uju6ob7rcnvd3mqcavkgc6ey

BASS: Boosting Abstractive Summarization with Unified Semantic Graph [article]

Wenhao Wu, Wei Li, Xinyan Xiao, Jiachen Liu, Ziqiang Cao, Sujian Li, Hua Wu, Haifeng Wang
2021 arXiv   pre-print
Further, a graph-based encoder-decoder model is proposed to improve both the document representation and summary generation process by leveraging the graph structure.  ...  Abstractive summarization for long-document or multi-document remains challenging for the Seq2Seq architecture, as Seq2Seq is not good at analyzing long-distance relations in text.  ...  Some other works also proposed to construct local knowledge graph by OpenIE to improve Seq2Seq models (Fan et al., 2019; Huang et al., 2020) .  ... 
arXiv:2105.12041v1 fatcat:rxo67twamrfsbgmb2fze6g7t5q

Template-Based Headline Generator for Multiple Documents

Yun-Chien Tseng, Mu-Hua Yang, Yao-Chung Fan, Wen-Chih Peng, Chih-Chieh Hung
2022 IEEE Access  
To the best of our knowledge, no one has used a technique for multi-document summarization to generate headlines in the past.  ...  The extractive stage is a graph-based model that identified salient sentences, whereas the abstractive stage uses existing summaries as soft templates to guild the seq2seq model.  ...  They would like to thank Yu-Chien Tang's help in experiments.  ... 
doi:10.1109/access.2022.3157287 fatcat:ad5c5huf3vhpzb4xa2z54fjtja

Incorporating Commonsense Knowledge into Story Ending Generation via Heterogeneous Graph Networks [article]

Jiaan Wang, Beiqi Zou, Zhixu Li, Jianfeng Qu, Pengpeng Zhao, An Liu, Lei Zhao
2022 arXiv   pre-print
In this paper, we propose a Story Heterogeneous Graph Network (SHGN) to explicitly model both the information of story context at different granularity levels and the multi-grained interactive relations  ...  In detail, we consider commonsense knowledge, words and sentences as three types of nodes. To aggregate non-local information, a global node is also introduced.  ...  We would like to thank Duo Zheng and the members of the NLPTTG group, Netease Fuxi AI Lab for the helpful discussions and valuable feedback.  ... 
arXiv:2201.12538v1 fatcat:4xk3j4s6lve6pkh4cg3tasnd2u

A Survey of Knowledge-Enhanced Text Generation [article]

Wenhao Yu, Chenguang Zhu, Zaitang Li, Zhiting Hu, Qingyun Wang, Heng Ji, Meng Jiang
2022 arXiv   pre-print
Since 2014, various neural encoder-decoder models pioneered by Seq2Seq have been proposed to achieve the goal by learning to map input text to output text.  ...  To address this issue, researchers have considered incorporating various forms of knowledge beyond the input text into the generation models.  ...  Fan et al. propose to construct an internal KG under a multiple input document scenario [29] .  ... 
arXiv:2010.04389v3 fatcat:vzdtlz4j65g2va7gwkbmzyxkhq

Multi-Hop Question Generation Using Hierarchical Encoding-Decoding and Context Switch Mechanism

Tianbo Ji, Chenyang Lyu, Zhichao Cao, Peng Cheng
2021 Entropy  
However, neural generation models suffer from the global and local semantic semantic drift problems.  ...  In the encoder, we hierarchically encode the input passages according to its structure at four granularity-levels: [word, chunk, sentence, document]-level.  ...  Acknowledgments: We would like to thank the anonymous crowd-sourcing raters for their work. Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/e23111449 pmid:34828147 pmcid:PMC8618393 fatcat:vl5bvrybq5glbnw6c7lty2sdpu

Incorporating Extra Knowledge to Enhance Word Embedding

Arpita Roy, Shimei Pan
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
In this survey, we summarize the recent advances in incorporating extra knowledge to enhance word embedding.  ...  Word embedding, a process to automatically learn the mathematical representations of words from unlabeled text corpora, has gained a lot of attention recently.  ...  Acknowledgments We would like to thank the anonymous reviewers for their constructive comments.  ... 
doi:10.24963/ijcai.2020/676 dblp:conf/ijcai/GaoCR0020 fatcat:n3hj4lad2vcphpmzdnwgflp7x4

From Standard Summarization to New Tasks and Beyond: Summarization with Manifold Information [article]

Shen Gao, Xiuying Chen, Zhaochun Ren, Dongyan Zhao, Rui Yan
2020 arXiv   pre-print
In general, text summarization algorithms aim at using a plain text document as input and then output a summary. However, in real-world applications, most of the data is not in a plain text format.  ...  Instead, there is much manifold information to be summarized, such as the summary for a web page based on a query in the search engine, extreme long document (e.g., academic paper), dialog history and  ...  Acknowledgements We would like to thank the anonymous reviewers for their constructive comments.  ... 
arXiv:2005.04684v1 fatcat:35ub2qoaezdq7fw7ptbvrbj37i

Leveraging Graph to Improve Abstractive Multi-Document Summarization [article]

Wei Li, Xinyan Xiao, Jiachen Liu, Hua Wu, Haifeng Wang, Junping Du
2020 arXiv   pre-print
In this paper, we develop a neural abstractive multi-document summarization (MDS) model which can leverage well-known graph representations of documents such as similarity graph and discourse graph, to  ...  Our model utilizes graphs to encode documents in order to capture cross-document relations, which is crucial to summarizing long documents.  ...  Introduction Multi-document summarization (MDS) brings great challenges to the widely used sequence-tosequence (Seq2Seq) neural architecture as it requires effective representation of multiple input documents  ... 
arXiv:2005.10043v1 fatcat:rne3yxooobbwpbrbimmlyzkfqu

Fine Grained Named Entity Recognition via Seq2seq Framework

Huiming Zhu, Chunhui He, Yang Fang, Weidong Xiao
2020 IEEE Access  
Fine-grained Named entity recognition (NER) is crucial to natural language processing (NLP) applications like relation extraction and knowledge graph construction.  ...  Such tagging scheme transfers NER problem into a sequence-to-sequence (seq2seq) based issue. We propose a seq2seq framework to comprehend the input sentence in a comprehensive way.  ...  graph construction.  ... 
doi:10.1109/access.2020.2980431 fatcat:fr5axng35rhj5jtsjeml4wzcxa

Flight Demand Forecasting with Transformers [article]

Liya Wang, Amy Mykityshyn, Craig Johnson, Jillian Cheng
2021 arXiv   pre-print
Transformers can enable artificial intelligence (AI) models to dynamically focus on certain parts of their input and thus reason more effectively.  ...  We then trained forecasting models with temporal fusion transformer (TFT) for five different airports.  ...  Government under Contract DTFAWA-10-C-00080 and is subject to Federal Aviation Administration Acquisition Management System Clause 3.5-13, Rights in Data-General, Alt. III and Alt. IV (Oct. 1996).  ... 
arXiv:2111.04471v1 fatcat:qjh6ggnyt5dvvcqlogx3abyqoy

Structure Learning for Headline Generation

Ruqing Zhang, Jiafeng Guo, Yixing Fan, Yanyan Lan, Xueqi Cheng
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
However, since documents are typically weakly-structured data, modern graph-based neural models usually rely on manually designed rules or some heuristics to construct the sentence graph a prior.  ...  That is, we want to automatically learn the sentence graph using a data-driven way, so that we can unveil the document structure flexibly without prior heuristics or rules.  ...  Graph Convolution Network After constructing the sentence graph, we feed the graph into a simple multi-layer Graph Convolutional Network (GCN) as in (Kipf and Welling 2017) .  ... 
doi:10.1609/aaai.v34i05.6501 fatcat:p3mmefn6hvhepbexeaqojcyoc4

Graph Neural Networks for Natural Language Processing: A Survey [article]

Lingfei Wu, Yu Chen, Kai Shen, Xiaojie Guo, Hanning Gao, Shucheng Li, Jian Pei, Bo Long
2021 arXiv   pre-print
To the best of our knowledge, this is the first comprehensive overview of Graph NeuralNetworks for Natural Language Processing.  ...  models.  ...  to the natural question given a large scale of open-domain knowledge (e.g. documents, knowledge base and etc.).  ... 
arXiv:2106.06090v1 fatcat:zvkhinpcvzbmje4kjpwjs355qu

Triples-to-Text Generation with Reinforcement Learning Based Graph-augmented Neural Networks [article]

Hanning Gao, Lingfei Wu, Hongyun Zhang, Zhihua Wei, Po Hu, Fangli Xu, Bo Long
2022 arXiv   pre-print
Most previous methods solve this task using a sequence-to-sequence model or using a graph-based model to encode RDF triples and to generate a text sequence.  ...  To solve these problems, we propose a model combining two new graph-augmented structural neural encoders to jointly learn both local and global structural information in the input RDF triples.  ...  First, the graph construction module leverages the input RDF triples to build two input graphs.  ... 
arXiv:2111.10545v3 fatcat:pcjq6ow2evaczaqzhiisn5etru
« Previous Showing results 1 — 15 out of 328 results