Neural Relation Extraction on Wikipedia Tables for Augmenting Knowledge Graphs

Macdonald, Erin M.; Barbosa, Denilson

doi:10.1145/3340531.3412164

Cited by 15 publications

(10 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides Wikipedia, we want to apply the approach to wikis in the Fandom 17 universe containing more than 380k wikis on various domains (among them many interesting wikis for our approach, like for example WikiLists 18 ). For background knowledge, we plan to rely on existing knowledge graphs in this domain like DBkWik [12] or TiFi [3].…”

Section: Discussion and Outlookmentioning

confidence: 99%

“…Oulabi and Bizer [21] tackle the same problem for Wikipedia tables with a bootstrapping approach based on expert-defined rules. Macdonald and Barbosa [17] extract new facts from Wikipedia tables to extend the Freebase knowledge base. With an LSTM that uses contextual information of the table, they extract new facts for 28 relations.…”

Section: Knowledge Graph Completion From Listingsmentioning

confidence: 99%

“…The previously mentioned approach of Macdonald and Barbosa [17] focuses on tables in Wikipedia and hence uses specific context features like section titles, table headers and captions, and the text in the first paragraph of the table's section. Interestingly, they do not only discover relations between entities in the table, but also between a table entity and the page subject.…”

Section: Exploiting the Context Of Listingsmentioning

confidence: 99%

See 2 more Smart Citations

Information Extraction From Co-Occurring Similar Entities

Heist

Paulheim

2021

Proceedings of the Web Conference 2021

View full text Add to dashboard Cite

Knowledge about entities and their interrelations is a crucial factor of success for tasks like question answering or text summarization. Publicly available knowledge graphs like Wikidata or DBpedia are, however, far from being complete. In this paper, we explore how information extracted from similar entities that co-occur in structures like tables or lists can help to increase the coverage of such knowledge graphs. In contrast to existing approaches, we do not focus on relationships within a listing (e.g., between two entities in a table row) but on the relationship between a listing's subject entities and the context of the listing. To that end, we propose a descriptive rule mining approach that uses distant supervision to derive rules for these relationships based on a listing's context. Extracted from a suitable data corpus, the rules can be used to extend a knowledge graph with novel entities and assertions. In our experiments we demonstrate that the approach is able to extract up to 3M novel entities and 30M additional assertions from listings in Wikipedia. We find that the extracted information is of high quality and thus suitable to extend Wikipedia-based knowledge graphs like DBpedia, YAGO, and CaLiGraph. For the case of DBpedia, this would result in an increase of covered entities by roughly 50%.

show abstract

Section: Discussion and Outlookmentioning

confidence: 99%

Section: Knowledge Graph Completion From Listingsmentioning

confidence: 99%

Section: Exploiting the Context Of Listingsmentioning

confidence: 99%

See 1 more Smart Citation

Information Extraction From Co-Occurring Similar Entities

Heist

Paulheim

2021

Proceedings of the Web Conference 2021

View full text Add to dashboard Cite

show abstract

“…While those techniques focus on identifying the semantic types of individual columns, there is another line of work that focused on column relations between pairs of columns in the same table for better understanding tables [4,12,23,24,28,46]. A column relation is a semantic label between a pair of columns in a table, which offers more fine-grained information about the table.…”

Section: Related Workmentioning

confidence: 99%

Annotating Columns with Pre-trained Language Models

Suhara,

Li,

et al. 2021

Preprint

View full text Add to dashboard Cite

Inferring meta information about tables, such as column headers or relationships between columns, is an active research topic in data management as we find many tables are missing some of these information. In this paper, we study the problem of annotating table columns (i.e., predicting column types and the relationships between columns) using only information from the table itself. We show that a multi-task learning approach (called Doduo), trained using pre-trained language models on both tasks outperforms individual learning approaches. Experimental results show that Doduo establishes new state-of-the-art performance on two benchmarks for the column type prediction and column relation prediction tasks with up to 4.0% and 11.9% improvements, respectively. We also establish that Doduo can already perform the previous stateof-the-art performance with a minimal number of tokens, only 8 tokens per column.

show abstract

“…Then, word combinations with the extracted partof-speech compositions were extracted. As data accumulates on wikipedia, various studies using the data are being conducted [15], [16], [17]. In this study, it was used to determine whether the text extracted through wikiapi is an existing term.After examining whether there was a Wikipedia page for the extracted word combination using wikiapi, the corresponding terms were extracted.…”

Section: Extraction Of Terms Other Than the Learning Elements In Textbooksmentioning

confidence: 99%

Relationship between Korean Informatics Curriculum and Textbook Learning Element Considering Compound Word

Kim

Woo

Kim

et al. 2021

JOIV : Int. J. Inform. Visualization

View full text Add to dashboard Cite

With the development of information and communication technology, countries around the world have strengthened their computer science curriculums. Korea also revised the informatics curriculum(The name of a subject related to computer science in Korea is informatics.) in 2015 with a focus on computer science. The purpose of this study was to automatically extract and analyze whether textbooks reflected the learning elements of the informatics curriculum in South Korea. Considering the forms of terms of the learning elements mainly comprised of compound words and the characteristics of Korean language, which makes natural language processing difficult due to various transformations, this study pre-processed textbook texts and the learning elements and derived their reflection status and frequencies. The terms used in the textbooks were automatically extracted by using the indexes in the textbooks and the part-of-speech compositions of the indexes. Moreover, this study analyzed the relevance between the terms by deriving confidence of other terms for each learning element used in the textbooks. As a result of the analysis, this study revealed that the textbooks did not reflect some learning elements in the forms presented in the curriculum, suggesting that the textbooks need to explain the concepts of the learning elements by using the forms presented in the curriculum at least once. This study is meaningful in that terms were automatically extracted and analyzed in Korean textbooks based on the words suggested by the curriculum. Also, the method can be applied equally to textbooks of other subjects.

show abstract

Neural Relation Extraction on Wikipedia Tables for Augmenting Knowledge Graphs

Cited by 15 publications

References 21 publications

Information Extraction From Co-Occurring Similar Entities

Information Extraction From Co-Occurring Similar Entities

Annotating Columns with Pre-trained Language Models

Relationship between Korean Informatics Curriculum and Textbook Learning Element Considering Compound Word

Contact Info

Product

Resources

About