Facts That Matter

Ponza, Marco; Corro, Luciano Del; Weikum, Gerhard

doi:10.18653/v1/d18-1129

Cited by 12 publications

(9 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As we desire a limited number of only 5 key facts per story, we take advantage of the SalIE framework (Ponza et al, 2018) that aims to rate saliency for extracted facts, and use the derived facts with the highest saliency scores as the key facts. Using these frameworks, described in Technical Details below, we derive a training data set of 17 thousand sets of key-facts and their correlating stories.…”

Section: Training Setmentioning

confidence: 99%

“…Saliency: Ponza et al (2018) built upon these frameworks defining an objective to evaluate the facts' salience scores, determining how essential is its information to the message the text conveys. Their framework, SalIE, rates facts extracted by MinIE and aims to output the most salient ones while maintaining sufficient diversity.…”

Section: Training Setmentioning

confidence: 99%

“…As left-to-right generation (GPT2, BART) tend Figure 1: Training sample for the Facts-to-text task. The facts were automatically extracted from the plot of "Scarface" (1932) using the SalIE system (Ponza et al, 2018). Note that a fact does not correspond to a contiguous span in the story.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Facts2Story: Controlling Text Generation by Key Facts

Orbach¹,

Goldberg²

2020

Proceedings of the 28th International Conference on Computational Linguistics

View full text Add to dashboard Cite

Recent advancements in self-attention neural network architectures have raised the bar for openended text generation. Yet, while current methods are capable of producing a coherent text which is several hundred words long, attaining control over the content that is being generated-as well as evaluating it-are still open questions. We propose a controlled generation task which is based on expanding a sequence of facts, expressed in natural language, into a longer narrative. We introduce human-based evaluation metrics for this task, as well as a method for deriving a large training dataset. We evaluate three methods on this task, based on fine-tuning pre-trained models. We show that while auto-regressive, unidirectional Language Models such as GPT2 produce better fluency, they struggle to adhere to the requested facts. We propose a plan-andcloze model (using fine-tuned XLNet) which produces competitive fluency while adhering to the requested content.

show abstract

Section: Training Setmentioning

confidence: 99%

Section: Training Setmentioning

confidence: 99%

See 1 more Smart Citation

Facts2Story: Controlling Text Generation by Key Facts

Orbach¹,

Goldberg²

2020

Proceedings of the 28th International Conference on Computational Linguistics

View full text Add to dashboard Cite

show abstract

“…Open Information Extraction (OpenIE) is an ontology-free information extraction paradigm that generates extractions of the form (subject; relation; object). Built on the principles of domainindependence and scalability (Mausam, 2016), OpenIE systems extract open relations and arguments from the sentence, which allow them to be *Equal Contribution 1 https://github.com/dair-iitd/openie6 used for a wide variety of downstream tasks like Question Answering (Yan et al, 2018;Khot et al, 2017), Event Schema Induction (Balasubramanian et al, 2013) and Fact Salience (Ponza et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

OpenIE6: Iterative Grid Labeling and Coordination Analysis for Open Information Extraction

Kolluru¹,

Adlakha²,

Aggarwal³

et al. 2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

View full text Add to dashboard Cite

A recent state-of-the-art neural open information extraction (OpenIE) system generates extractions iteratively, requiring repeated encoding of partial outputs. This comes at a significant computational cost. On the other hand, sequence labeling approaches for OpenIE are much faster, but worse in extraction quality. In this paper, we bridge this trade-off by presenting an iterative labeling-based system that establishes a new state of the art for OpenIE, while extracting 10× faster. This is achieved through a novel Iterative Grid Labeling (IGL) architecture, which treats OpenIE as a 2-D grid labeling task. We improve its performance further by applying coverage (soft) constraints on the grid at training time.Moreover, on observing that the best OpenIE systems falter at handling coordination structures, our OpenIE system also incorporates a new coordination analyzer built with the same IGL architecture. This IGL based coordination analyzer helps our OpenIE system handle complicated coordination structures, while also establishing a new state of the art on the task of coordination analysis, with a 12.3 pts improvement in F1 over previous analyzers. Our OpenIE system, OpenIE6 1 , beats the previous systems by as much as 4 pts in F1, while being much faster.

show abstract

“…Detecting salient information in documents, such as sentences, open facts, keywords, or Wikipedia entities, has become a fundamental task on which different information retrieval (IR) and Natural Language Processing (NLP) tools hinge upon to improve their performance. Contextual ads‐matching, document similarity, web search ranking, and news suggestion are just a few examples of typical research domains on which the salient information extracted from natural language texts is consumed.…”

Section: Introductionmentioning

confidence: 99%

Swat: A system for detecting salient Wikipedia entities in texts

Ponza

Ferragina

Piccinno

2019

Computational Intelligence

Self Cite

View full text Add to dashboard Cite

We study the problem of entity salience by proposing the design and implementation of Swat, a system that identifies the salient Wikipedia entities occurring in an input document. Swat consists of several modules that are able to detect and classify on‐the‐fly Wikipedia entities as salient or not, based on a large number of syntactic, semantic, and latent features properly extracted via a supervised process, which has been trained over millions of examples drawn from the New York Times corpus. The validation process is performed through a large experimental assessment, eventually showing that Swat improves known solutions over all publicly available datasets. We release Swat via an API that we describe and comment in the paper to ease its use in other software.

show abstract

Facts That Matter

Cited by 12 publications

References 20 publications

Facts2Story: Controlling Text Generation by Key Facts

Facts2Story: Controlling Text Generation by Key Facts

OpenIE6: Iterative Grid Labeling and Coordination Analysis for Open Information Extraction

Swat: A system for detecting salient Wikipedia entities in texts

Contact Info

Product

Resources

About