Coreference Resolution: Toward End-to-End and Cross-Lingual Systems

Cruz, André Ferreira; Rocha, Gil; Cardoso, Henrique Lopes

doi:10.3390/info11020074

Cited by 9 publications

(3 citation statements)

References 86 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the case of DIAVGEIA, GreekBERT (Koutsikakis et al, 2020) provides a perfect resource for this domain along with Spacy's pre-trained NER and CP models. For the remaining tasks, utilizing multilingual models provide the desired functionality (Cruz et al, 2020;De Cao et al, 2021). Overall, multilingual models have shown significant improvement in the latest research (Baumann, 2019;Gromann, 2020).…”

Section: Discussionmentioning

confidence: 99%

Doc2KG

Vlachava

Konstantinidis

Bassiliades

et al. 2022

International Journal on Semantic Web and Information Systems

View full text Add to dashboard Cite

Document Management Systems (DMS) are used for decades to store large amounts of information in textual form. Their technology paradigm is based on storing vast quantities of textual information enriched with metadata to support searchability. However, this exhibits limitations as it treats textual information as black box and is based exclusively on user-created metadata, a process that suffers from quality and completeness shortcomings. The use of knowledge graphs in DMS can substantially improve searchability, providing the ability to link data and enabling semantic searching. Recent approaches focus on either creating knowledge graphs from document collections or updating existing ones. In this paper, we introduce Doc2KG (Document-to-Knowledge-Graph), an intelligent framework that handles both creation and real-time updating of a knowledge graph, while also exploiting domain-specific ontology standards. We use DIAVGEIA (clarity), an award winning Greek open government portal, as our case-study and discuss new capabilities for the portal by implementing Doc2KG.

show abstract

Section: Discussionmentioning

confidence: 99%

Doc2KG

Vlachava

Konstantinidis

Bassiliades

et al. 2022

International Journal on Semantic Web and Information Systems

View full text Add to dashboard Cite

show abstract

“…Automatic neural NER systems have achieved accuracy exceeding 92% F 1 scores in many languages and domains (Wang et al, 2019;Akbik et al, 2018). NER tags produced by these systems are useful in many other natural language processing contexts, such as coreference resolution, entity linking, or entity extraction (Ferreira Cruz et al, 2020). POS taggers have also achieved very high accuracy exceeding 98% on public treebank datasets (Akbik et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

Tag Assisted Neural Machine Translation of Film Subtitles

Siekmeier¹,

Lee²,

Kwon³

et al. 2021

Proceedings of the 18th International Conference on Spoken Language Translation (IWSLT 2021)

View full text Add to dashboard Cite

We implemented a neural machine translation system that uses automatic sequence tagging to improve the quality of translation. Instead of operating on unannotated sentence pairs, our system uses pre-trained tagging systems to add linguistic features to source and target sentences. Our proposed neural architecture learns a combined embedding of tokens and tags in the encoder, and simultaneous token and tag prediction in the decoder. Compared to a baseline with unannotated training, this architecture increased the BLEU score of German to English film subtitle translation outputs by 1.61 points using named entity tags; however, the BLEU score decreased by 0.38 points using part-of-speech tags. This demonstrates that certain token-level tag outputs from off-theshelf tagging systems can improve the output of neural translation systems using our combined embedding and simultaneous decoding extensions.

show abstract

“…Attribution is a critical feature of journalism (Newell et al , 2017), a fundamental rhetorical mechanism in the media (Newell et al , 2018); it supports the journalist’s identified claims and viewpoints (Schneider et al , 2010) and provides transparency and accountability to readers by making identifiable news sources (Esser and Umbricht, 2014). Attribution is also broadly used from fact-checking to opinion mining (Cruz et al , 2020). Attribution aspects, such as sources, stances and in-depth information, can significantly affect how the information is perceived by readers (Pareti, 2012).…”

Section: Introductionmentioning

confidence: 99%

Understanding quotation extraction and attribution: towards automatic extraction of public figure’s statements for journalism in Indonesia

Purnomo

Kumar

Zulkarnain

2020

GKMC

View full text Add to dashboard Cite

Purpose Extracting information from unstructured data becomes a challenging task for computational linguistics. Public figure’s statement attributed by journalists in a story is one type of information that can be processed into structured data. Therefore, having the knowledge base about this data will be very beneficial for further use, such as for opinion mining, claim detection and fact-checking. This study aims to understand statement extraction tasks and the models that have already been applied to formulate a framework for further study. Design/methodology/approach This paper presents a literature review from selected previous research that specifically addresses the topics of quotation extraction and quotation attribution. Research works that discuss corpus development related to quotation extraction and quotation attribution are also considered. The findings of the review will be used as a basis for proposing a framework to direct further research. Findings There are three findings in this study. Firstly, the extraction process still consists of two main tasks, namely, the extraction of quotations and the attribution of quotations. Secondly, most extraction algorithms rely on a rule-based algorithm or traditional machine learning. And last, the availability of corpus, which is limited in quantity and depth. Based on these findings, a statement extraction framework for Indonesian language corpus and model development is proposed. Originality/value The paper serves as a guideline to formulate a framework for statement extraction based on the findings from the literature study. The proposed framework includes a corpus development in the Indonesian language and a model for public figure statement extraction. Furthermore, this study could be used as a reference to produce a similar framework for other languages.

show abstract

Coreference Resolution: Toward End-to-End and Cross-Lingual Systems

Cited by 9 publications

References 86 publications

Doc2KG

Doc2KG

Tag Assisted Neural Machine Translation of Film Subtitles

Understanding quotation extraction and attribution: towards automatic extraction of public figure’s statements for journalism in Indonesia

Contact Info

Product

Resources

About