Utilizing Out-Domain Datasets to Enhance Multi-task Citation Analysis

Mercier, Dominique; Rizvi, Syed Tahseen Raza; Rajashekar, Vikas; Ahmed, Sheraz; Dengel, Andreas

doi:10.1007/978-3-031-10161-8_6

Cited by 2 publications

(1 citation statement)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They achieved the macro-F1 of 88.93 and 77.73 for SciCite and CSC-C respectively. In an extension of prior research, author [27], conducted multi-task learning encompassing sentiment analysis and citation sentiment analysis. They utilized a modified XLNet to simultaneously learn distributions from features acquired from diverse sources such as movie and product reviews, Twitter data, and CSC-C. Training involved out-of-domain, sequential, and shuffled approaches across datasets from four domains.…”

Section: Literature Surveymentioning

confidence: 99%

Multi-sentence and multi-intent classification using RoBERTa and graph convolutional neural network

Ravi,

Singh,

Gautam

2024

Preprint

View full text Add to dashboard Cite

Citation analysis has garnered significant attention in academia, particularly in the realm of scientometrics analysis. Most studies related to citation analysis focus on quantitative aspects, assigning equal weight to every citation regardless of its placement within the paper. However, understanding the distribution of citation weight across different sections of a research article is crucial for citation analysis and impact assessment. Therefore, the analysis of citation intent becomes a pivotal task in determining the qualitative importance of a citation within a scientific article. In this context, we undertook two essential tasks related to citation analysis: citation length analysis and citation intent analysis. Through citation length analysis, we identified the optimal number of citation sentences to consider around a cited sentence. Simultaneously, citation intent analysis aimed to categorize citations into seven distinct types, namely background, motivation, uses, extends, similarities, differences, and future work. For the latter task, we introduced two novel architectures based on graph neural networks, namely CiteIntentRoBERTaGCN and CiteIntentRoBERTaGAT. The performance of these proposed models was evaluated on five multi-intent datasets curated from 1,200 research papers, considering different context lengths. The results demonstrated that the proposed models achieved state-of-the-art performance.

show abstract

Section: Literature Surveymentioning

confidence: 99%

Multi-sentence and multi-intent classification using RoBERTa and graph convolutional neural network

Ravi,

Singh,

Gautam

2024

Preprint

View full text Add to dashboard Cite

show abstract

ACE 2.0: A Comprehensive tool for automatic extraction, analysis, and digital profiling of the researchers in Scientific Communities

Rizvi

Ahmed

Dengel

2023

Soc. Netw. Anal. Min.

View full text Add to dashboard Cite

In the current digital era, it is remarkably convenient for researchers to share and collaborate on novel scientific ideas. Scientists aim to accomplish these endeavors through closely knitted scientific communities, depending on the domain. Technological advancements and their evolution overtime gave rise to a boom in the emergence of research communities with unique topics and focuses. Due to the enormous number and vastness of scientific communities, it is an intractable task to analyze scientific communities and administer them from a quantitative and qualitative perspective. Existing tools provide a limited and shallow glance into a scientific community. In this paper, we present a comprehensive system for the analysis of scientific communities called ACE 2.0 (Academic Community Explorer 2.0) which employs state-of-the-art models to automatically, efficiently, and smartly extract, and analyze bibliographic data. Moreover, it provides a range of insights from individual researchers to interactions between communities. These insights include different community-level aspects like collaboration patterns, citation patterns, influential persons with different roles, contributions from geographical locations, topics evolution, and many other fine-grained aspects within each scientific community. Our system considers scholarly publications as a primary source of information. However, it also employs several external resources to collect as much data as possible to correctly identify individual researchers and their contributions. Using all the collected data, ACE 2.0 performs an analysis of scientific communities and automatically performs detailed digital profiling of individual researchers. This analysis identifies trends in their citation, collaboration, contributions, popularity, and role in the community. Additionally, ACE 2.0 introduces a new Semantic index for researchers that takes into account both quantitative and qualitative aspects of the citations received by a researcher and quantifies their influence in the community. To conclude, ACE 2.0 enables us to analyze and oversee the scientific communities using trends and information gathered from different sources encompassing multiple aspects. Therefore, this work motivates us to discover endless new perspectives and opens it up to a wide range of applications in other domains. The demo of ACE 2.0 visualization engine is available at https://ace.opendfki.de/.

show abstract

Utilizing Out-Domain Datasets to Enhance Multi-task Citation Analysis

Cited by 2 publications

References 26 publications

Multi-sentence and multi-intent classification using RoBERTa and graph convolutional neural network

Multi-sentence and multi-intent classification using RoBERTa and graph convolutional neural network

ACE 2.0: A Comprehensive tool for automatic extraction, analysis, and digital profiling of the researchers in Scientific Communities

Contact Info

Product

Resources

About