BLAB Reporter Automated journalism covering the Blue Amazon Yan Vianna Sym Escola Politécnica

2025-04-27 0 0 154.23KB 4 页 10玖币

侵权投诉

BLAB Reporter: Automated journalism covering the Blue Amazon

Yan Vianna Sym

Escola Politécnica

Universidade de São Paulo

São Paulo, Brazil

yan.sym@usp.br

João Gabriel Moura Campos

Escola Politécnica

Universidade de São Paulo

São Paulo, Brazil

joaogcampos@usp.br

Fabio Gagliardi Cozman

Escola Politécnica

Universidade de São Paulo

São Paulo, Brazil

fgcozman@usp.br

Abstract

This demo paper introduces the BLAB Re-

porter, a robot-journalist covering the Brazil-

ian Blue Amazon. The Reporter is based

on a pipeline architecture for Natural Lan-

guage Generation; it offers daily reports,

news summaries and curious facts in Brazil-

ian Portuguese. By collecting, storing and

analysing structured data from publicly avail-

able sources, the robot-journalist uses domain

knowledge to generate and publish texts in

Twitter. Code and corpus are publicly avail-

able 1.

1 Introduction

Data-to-text Natural Language Generation (NLG)

is the computational process of generating mean-

ingful and coherent natural text or speech to de-

scribe non-linguistic input data (Reiter and Dale,

2000). Successful examples of data-to-text systems

can be found in both academia and industry, with

applications in weather forecasting (Belz,2008),

image captions and chatbots (Adamopoulou and

Moussiades,2020). Amongst NLG applications,

robot-journalism is one of the most prominent en-

deavors thanks to the high volume of structured

data streams available, which enables automated

systems to report recurrent information with high-

ﬁdelity and lexical variety (Teixeira et al.,2020).

An interesting domain for data-to-text genera-

tion is ocean monitoring. For instance, global at-

tention was drawn in 2021 to a container ship that

obstructed the Suez Canal for six consecutive days,

causing a global shortage of essential commodities,

including medical supplies and medicines during

the coronavirus pandemic. Accurate and low la-

tency information reports can be very helpful in

these situations, but communicating to general au-

diences usually demands coverage by specialized

human journalists. To address this issue, we present

1https://github.com/C4AI/blab-reporter

our robot-journalist named BLAB Reporter, a NLG

system based on a pipeline architecture that gen-

erates daily reports, news, content summarization

and curious facts about the Blue Amazon and pub-

lishes them on Twitter in Brazilian Portuguese

The Blue Amazon is the exclusive economic zone

(EEZ) of Brazil, with an offshore area of 3.6 mil-

lion square kilometers along the Brazilian coast,

an area rich in marine biodiversity and energy re-

sources (Wiesebron,2013). The BLue Amazon

Brain (BLAB) is a project aiming to address com-

plex questions about the marine ecosystem, and

integrates a number of services aimed at dissemi-

nating information about the Blue Amazon.

2 System overview

Our system follows a pipeline architecture that con-

verts non-linguistic data into text in 6 steps: Con-

tent Selection, Discourse Ordering, Text Structur-

ing, Lexicalization, Referring Expression Genera-

tion and Textual Realization (Ferreira et al.,2019).

Our system also comprises two additional steps:

Data Acquisition (for extracting and storing infor-

mation from multiple data streams in a structured

format) and Summarization (for summarizing news

in the form of small consecutive tweets). This kind

of architecture, depicted in Figure 1, allows for

trustworthy output as well as easy access to and

maintenance of sub-modules.

The grammar used by the model was built by

ﬁrst running the content selection step in previ-

ous data and generating 30 non-linguistic reports.

These non-linguistic reports were then manually

verbalized and the input and output representations

for each pipeline module were manually annotated.

When deployed, each module draws on the se-

lected combination of templates using rule-based

approaches. Because we deal with a sensitive do-

main, we opted to use the pipeline architecture

2https://twitter.com/BLAB_Reporter

arXiv:2210.06431v1 [cs.CL] 8 Oct 2022

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

BLABReporter:AutomatedjournalismcoveringtheBlueAmazonYanViannaSymEscolaPolitécnicaUniversidadedeSãoPauloSãoPaulo,Brazilyan.sym@usp.brJoãoGabrielMouraCamposEscolaPolitécnicaUniversidadedeSãoPauloSãoPaulo,Braziljoaogcampos@usp.brFabioGagliardiCozmanEscolaPolitécnicaUniversidadedeSãoPauloSãoPaulo,Brazi...

展开>> 收起<<

BLAB Reporter Automated journalism covering the Blue Amazon Yan Vianna Sym Escola Politécnica.pdf

共4页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

BLAB Reporter Automated journalism covering the Blue Amazon Yan Vianna Sym Escola Politécnica

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: