Improving Scholarly Knowledge Representation: Evaluating BERT-Based Models for Scientific Relation Classification

Ji, Meng; D’Souza, Jennifer; Auer, Sören; Downie, J. Stephen

doi:10.1007/978-3-030-64452-9_1

Cited by 7 publications

(5 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The analysis of the entire dataset shows that all the entries contain the title field; 87.11% of the entries contain both the title and the subject heading fields; 52.26% of the items contain both the title and the abstract fields; and 51.34% of the items contain the title, the subject heading, and the abstract fields. Therefore, it can be seen that for classified books, the title and the subject heading are relatively common fields and the abstract field has a relatively low probability (Jiang et al, 2020). This section will select the basic one-way LSTM model: one layer of LSTM hidden layer; each hidden layer contains 128 nodes; the amount of data per batch is 128; the training process adopts the principle of early stopping, when the loss value of the model on the validation set Stop training when it increases, and train the entire training set for up to 1,000 epochs.…”

Section: Exploring the Impact Of Field Selection On Classificationmentioning

confidence: 99%

Research on Library Resource Management Based on Modern Information Technology and Reconfigurable Mobile Information System

Zhang

2024

Journal of Cases on Information Technology

View full text Add to dashboard Cite

Library resource library management based on modern information technology and reconfigurable mobile information systems have attracted much attention from researchers of academia. This paper introduces the LSTM model into the field of Chinese library resource management, and tries to propose a classification system with practical application value. The main research work:The Chinese library resource is designed LSTM model, and three experiments are designed on this basis: the coarse-grained book single classification experiment, different measures on the classification results; the fine-grained book single classification experiment, to compare The advantages and disadvantages of direct classification and layer-by-layer classification; coarse-grained book multi-classification experiments to explore the feasibility of multi-classification. After analyzing the experimental results, the model has achieved a good classification effect, and overcomes some deficiencies in the library resource library predecessors, and has a strong practical application value.

show abstract

Section: Exploring the Impact Of Field Selection On Classificationmentioning

confidence: 99%

Research on Library Resource Management Based on Modern Information Technology and Reconfigurable Mobile Information System

Zhang

2024

Journal of Cases on Information Technology

View full text Add to dashboard Cite

show abstract

“…The structure is simply logical, with the goal of maximizing the reasoning in our scenario. [41] is explored to identify relation types in knowledge graphs in scholarly domain. Farber in [42] developed a framework for extracting entities such as scientific methods and data set along with classification and aggregation.…”

Section: Common Scholarly Communication Infrastructuresmentioning

confidence: 99%

Scholarly knowledge graphs through structuring scholarly communication: a review

Verma

Bhatia

Harit

et al. 2022

Complex Intell. Syst.

View full text Add to dashboard Cite

The necessity for scholarly knowledge mining and management has grown significantly as academic literature and its linkages to authors produce enormously. Information extraction, ontology matching, and accessing academic components with relations have become more critical than ever. Therefore, with the advancement of scientific literature, scholarly knowledge graphs have become critical to various applications where semantics can impart meanings to concepts. The objective of study is to report a literature review regarding knowledge graph construction, refinement and utilization in scholarly domain. Based on scholarly literature, the study presents a complete assessment of current state-of-the-art techniques. We presented an analytical methodology to investigate the existing status of scholarly knowledge graphs (SKG) by structuring scholarly communication. This review paper investigates the field of applying machine learning, rule-based learning, and natural language processing tools and approaches to construct SKG. It further presents the review of knowledge graph utilization and refinement to provide a view of current research efforts. In addition, we offer existing applications and challenges across the board in construction, refinement and utilization collectively. This research will help to identify frontier trends of SKG which will motivate future researchers to carry forward their work.

show abstract

“…As a pre-trained language representation built on the deep neural technology of transformers, it provides NLP practitioners with high-quality language features from text data simply out-ofthe-box and thus improves performance on many NLP tasks. These models return contextualized word embeddings that can be directly employed as features for downstream tasks [28].…”

Section: Bert Modelsmentioning

confidence: 99%

“…Prospectively, machine learning can assist scientists to record their results in the Leaderboards of next-generation digital libraries such as the Open Research Knowledge Graph (ORKG) [27]. In our age of the "deep learning tsunami," [38] there are many studies that have used neural network models to improve the construction of automated scholarly knowledge mining systems [36,12,7,28]. With the recent introduction of language modeling techniques such as transformers [44], the opportunity to obtain boosted machine learning systems is further accentuated.…”

Section: Introductionmentioning

confidence: 99%

“…Would it work in the real-world setting of large amounts of data? Thus, we note that it should be made possible to recommend a technique for knowledge organization services from observations based on our prior comprehensive empirical evaluations [28]. Our ultimate goal with this study is to help the DL stakeholders to select the optimal tool to implement knowledge-based scientific information flows w.r.t.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automated Mining of Leaderboards for Empirical AI Research

Kabongo¹,

D’Souza²,

Auer³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

With the rapid growth of research publications, empowering scientists to keep oversight over the scientific progress is of paramount importance. In this regard, the Leaderboards facet of information organization provides an overview on the state-of-the-art by aggregating empirical results from various studies addressing the same research challenge. Crowdsourcing efforts like PapersWithCode among others are devoted to the construction of Leaderboards predominantly for various subdomains in Artificial Intelligence. Leaderboards provide machine-readable scholarly knowledge that has proven to be directly useful for scientists to keep track of research progress. The construction of Leaderboards could be greatly expedited with automated text mining. This study presents a comprehensive approach for generating Leaderboards for knowledge-graph-based scholarly information organization. Specifically, we investigate the problem of automated Leaderboard construction using state-of-the-art transformer models, viz. Bert, SciBert, and XLNet. Our analysis reveals an optimal approach that significantly outperforms existing baselines for the task with evaluation scores above 90% in F1. This, in turn, offers new state-of-the-art results for Leaderboard extraction. As a result, a vast share of empirical AI research can be organized in the next-generation digital libraries as knowledge graphs.

show abstract

Improving Scholarly Knowledge Representation: Evaluating BERT-Based Models for Scientific Relation Classification

Cited by 7 publications

References 22 publications

Research on Library Resource Management Based on Modern Information Technology and Reconfigurable Mobile Information System

Research on Library Resource Management Based on Modern Information Technology and Reconfigurable Mobile Information System

Scholarly knowledge graphs through structuring scholarly communication: a review

Automated Mining of Leaderboards for Empirical AI Research

Contact Info

Product

Resources

About