How to use scite to help with your research (scite webinar, Sep 29, 2021)

scite
29 Sept 202155:46

TLDRThe webinar 'How to use scite to help with your research' from September 29, 2021, introduces the scite tool, designed to enhance the reliability of research by providing detailed citation context. Josh Nicholson, co-founder and CEO of scite, demonstrates how the platform can analyze articles, researchers, and the next generation of citations, offering a more nuanced understanding of research impact beyond traditional metrics. The presentation also covers the browser extension, reference check tool, and the ability to create custom dashboards for tracking topics of interest.

Takeaways

  • 📝 The webinar introduces 'Scite', a tool aimed at making research more reliable by providing deeper insights into how published articles are cited and interpreted in the academic community.
  • 🔍 Scite offers a browser extension that can display citation information virtually anywhere a research article is read online, enhancing traditional citation metrics with context.
  • 💡 Scite uses machine learning to classify citation statements into supporting, mentioning, or contrasting evidence, providing a nuanced view of how research is referenced.
  • 🀖 The platform has partnered with over a dozen publishers to access content that is not freely available, utilizing tools like GROBID for text extraction and deep learning models for citation analysis.
  • 🔗 Scite provides a detailed view of citation contexts, allowing users to understand not just the quantity of citations but the quality and nature of the references made to a piece of work.
  • 📈 The tool offers various dashboards for articles, researchers, journals, affiliations, and funders, giving a comprehensive overview of citation patterns and impact.
  • 🗂 Scite's 'Reference Check' feature helps authors, editors, and peer reviewers analyze the references of a manuscript, ensuring the cited works are current and credible.
  • 🔒 Scite is careful to avoid false positives in its self-citation detection, taking a conservative approach to ensure accuracy in its analysis.
  • 📚 The platform has indexed a vast number of citation statements from millions of full-text articles, providing a rich dataset for users to explore.
  • 🔄 Scite acknowledges the foundational work of others in the field, such as Eugene Garfield, and builds upon these ideas to create a modern tool for citation analysis.
  • 🌐 The webinar concludes with an invitation for feedback and questions, emphasizing Scite's commitment to user engagement and continuous improvement.

Q & A

  • What is the purpose of the Scite tool as described in the webinar?

    -Scite is designed to make research more reliable by providing a deeper understanding of how scholarly articles have been cited, offering insights beyond just citation counts. It classifies citations as supporting, mentioning, or contrasting, allowing researchers to better interpret and utilize the cited work.

  • How does Scite differentiate between different types of citations?

    -Scite uses a deep learning model to analyze citation statements and classify them as supporting, mentioning, or contrasting evidence. This classification helps users understand the context and sentiment of the citations more accurately.

  • What is the significance of showing the citation context rather than just the citation count?

    -Showing the citation context provides a more nuanced understanding of how a piece of research is being used and interpreted by others in the field. It goes beyond the quantity of citations to reveal the quality and relevance of the citations.

  • How does Scite handle the full-text articles to extract citation information?

    -Scite utilizes open access sources and partnerships with publishers to access full-text articles. It uses an open-source tool called Grobid, which involves machine learning to extract in-text citations and the surrounding text for analysis.

  • What is the Scite browser extension, and how does it work?

    -The Scite browser extension is available for Chrome, Firefox, Safari, and Microsoft Edge. It shows Scite's citation information virtually anywhere a research article is read online, providing the number of paper citations and a breakdown of citation statements for each reference.

  • How does Scite classify citation statements, and what is the basis for this classification?

    -Scite trains a deep learning model on a dataset of around 50,000 citation statements that have been manually annotated by experts. The model then uses this training to classify new citation statements as supporting, mentioning, or contrasting.

  • What is the Scite Index, and how is it used to evaluate journals?

    -The Scite Index is a metric that measures the ratio of supporting citations to the total number of supporting plus contrasting citations for a given journal. It provides a complementary view to the impact factor, offering insight into the overall sentiment of citations received by the journal.

  • How can Scite be used to analyze the citation patterns of researchers or institutions?

    -Scite provides researcher and affiliation dashboards that show how an individual or institution has been cited, including the number of supporting, mentioning, and contrasting citations. This can help understand the impact and influence of their research within the academic community.

  • What is the Reference Check tool, and how does it assist authors and editors?

    -The Reference Check tool allows users to upload a document and automatically identifies the in-text citations, matches them to references, and checks them against Scite's database for any editorial concerns or citation patterns. This helps ensure that the references in a manuscript are reliable and relevant.

  • How does Scite ensure the accuracy of its citation classifications, and is there a mechanism for user feedback?

    -Scite's deep learning model is trained on a diverse dataset and tested against a holdout set of manually annotated citation statements. Users can flag citation statements they believe are misclassified, which are then reviewed by experts. This feedback mechanism helps improve the model's accuracy over time.

Outlines

00:00

📜 Introduction and Webinar Objectives

Josh Nicholson, the co-founder and CEO of Site, introduces the webinar aimed at demonstrating how Site can enhance research reliability and interpretation. He emphasizes the importance of scientific research and its impact on various aspects of life. Josh invites participants to interact by asking questions or providing feedback during the presentation. He also mentions Dom, who is available to answer technical questions. A special discount for the webinar attendees is provided, and Josh offers his contact information for further inquiries. The session begins with a reflection on the broader context of scientific publishing and the challenges of information overload and reproducibility in research.

05:02

🔍 Exploring Site's Citation Analysis Features

This paragraph delves into the functionality of Site, highlighting its ability to analyze how individual articles and researchers have been cited. It discusses the browser extension that provides citation information while reading research articles online. The extension shows the number of paper citations and breaks down citation statements into supporting, mentioning, and contrasting evidence. The paragraph also explains how Site uses machine learning to classify citation statements and the importance of full-text articles for extracting in-text citations. Site's partnerships with publishers to access content and the use of the Grobid tool for machine learning models are also covered.

10:03

🀖 Machine Learning in Citation Context Analysis

The speaker discusses the process of training a deep learning model to classify citation statements into supporting, mentioning, or contrasting categories. The model was trained on approximately 50,000 citation statements annotated by experts with MDs or PhDs from diverse disciplines. A holdout set of 10,000 citation statements is used to test the model's precision and recall. The challenges of accurately classifying citations due to variations in scientific writing and notation are acknowledged. The paragraph also covers how Site organizes citation information, allowing users to filter and explore citations based on various criteria such as year, paper section, and citation type.

15:03

📈 Visualization and Analysis of Citation Networks

Josh introduces the visualization of citation networks on Site, which helps users understand the relationships between cited works. The size of the nodes in the network visualization represents the number of citations a paper has received. Users can expand the network to see multiple papers and filter citations to focus on supporting evidence or contrasting views. The paragraph also discusses the potential for further development in visualization tools to improve the understanding of citation contexts and networks.

20:03

🏫 Insights into Journal and Affiliation Dashboards

This section covers the dashboards available for journals and affiliations on Site, providing insights into citation patterns, editorial notices, and the overall Site index, which is a ratio of supporting to contrasting citations. The dashboards offer a way to quickly assess the impact and reception of publications within specific journals or from certain affiliations. The paragraph also mentions the potential for creating subject-specific dashboards to further analyze citation trends in particular fields of study.

25:04

🔎 Advanced Search and Reference Check Tools

Josh presents advanced search capabilities on Site, including the ability to search for topics across papers and to filter results based on citation metrics. He introduces the Reference Check tool, designed to help authors, editors, and peer reviewers analyze the references of a manuscript. This tool can identify if cited references have been retracted or if they have been cited with supporting or contrasting evidence. The integration of Reference Check with major submission systems is discussed to automate the process of checking references.

30:05

🌐 Broadening the Scope: Social Sciences and Humanities

The webinar addresses the representation of social sciences and humanities on Site compared to experimental sciences. While the value of citation context is consistent across disciplines, the incidence of supporting or contrasting citations may vary. An analysis of the Site journal index across different subject areas shows a distribution that reflects the nature of different scientific fields and how researchers cite work within them. The paragraph emphasizes the importance of understanding these variations and the potential for further research in this area.

35:07

📚 Access to Full-Text Publications and New Citation Statement Search

This paragraph discusses the importance of access to full-text publications for citation analysis and acknowledges that Site does not have access to all content. It introduces a new citation statement search feature that allows users to search the entire database for expert analyses and opinions on virtually any topic. The feature is designed to provide quick insights into how topics have been discussed and cited in the research literature, enhancing the accessibility and understanding of research across various fields.

40:08

📘 Conclusion and Acknowledging Precedents

Josh concludes the webinar by acknowledging the pioneers in citation analysis, such as Eugene Garfield, and their foundational work that has inspired the development of Site. He emphasizes the importance of building upon previous research and ideas. The conclusion also highlights the potential for Site to improve and evolve based on user feedback and the team's commitment to listening and making prompt enhancements.

45:10

🀝 Final Q&A and Future Enhancements

The final paragraph covers the Q&A session where Josh and Dom address questions from participants about Site's capabilities, such as analyzing institutions and the validation of the supporting contrasting classification. They discuss the possibility of filtering institutions by topic and the ongoing process of validation and feedback from the research community. The paragraph ends with an invitation for further questions and feedback via email and a thank you to the participants for their engagement.

Mindmap

Keywords

💡scite

Scite is a tool designed to enhance the reliability and interpretability of research by providing a deeper understanding of how scholarly articles are cited. It uses machine learning to classify citation statements into supporting, mentioning, or contrasting evidence. In the video, scite is presented as a means to transform traditional citation metrics into a more nuanced view of a paper's impact and reception within the academic community.

💡reproducibility

Reproducibility in research refers to the ability of other scientists to recreate the results of an experiment or study. It is a cornerstone of the scientific method and ensures the validity of research findings. The script mentions challenges with reproducibility, indicating a need for tools like scite that can help assess the reliability of research by analyzing how it is cited and discussed in the literature.

💡citation context

Citation context is the specific part of a text that explains why and how a reference is being cited. In the script, scite's ability to provide citation context is highlighted as a way to move beyond mere citation counts to understand the nature of the citations, whether they support, mention, or contrast with the cited work.

💡deep learning model

A deep learning model is a type of artificial intelligence that can process and learn from large amounts of data. In the context of the video, scite uses a deep learning model to classify citation statements at scale, which is crucial for its function in analyzing and providing insights into citation patterns across millions of articles.

💡bibliometrics

Bibliometrics is the application of statistical methods to books and other media to quantify and evaluate their impact and usage. The video discusses how scite goes beyond traditional bibliometrics by providing a more detailed analysis of citation statements, offering a richer understanding of a work's influence within its field.

💡browser extension

A browser extension is a software component that adds functionality to a web browser. The script mentions scite's browser extension, which allows users to access citation information virtually anywhere they are reading a research article online, enhancing the accessibility of the tool and its features.

💡reference check

Reference check is a tool within scite that helps authors, editors, and peer reviewers analyze the references of a manuscript. The video script describes how this tool can identify if cited works have been retracted or have been cited in ways that support or contrast with the original claims, thereby ensuring the integrity of the research being published.

💡editorial notices

Editorial notices are statements published alongside a research article that alert readers to issues such as retractions, corrections, or expressions of concern. The video script discusses how scite includes information about editorial notices, which is important for understanding the status and reliability of cited research.

💡site index

The site index is a metric used by scite to measure the overall supportiveness of citations received by a journal, article, or author. It is calculated as the ratio of supporting citations to the total number of supporting and contrasting citations. The script mentions the site index as a way to provide a complementary view on the impact and reception of research beyond traditional impact factors.

💡citation statement search

Citation statement search is a feature of scite that allows users to search the entire database of citation statements to find expert analyses, opinions, and interpretations on various topics. The video script presents this search function as a powerful tool for quickly accessing the research literature's stance on specific subjects, making research more accessible and understandable.

Highlights

Scite is a tool aimed at making research more reliable by providing a deeper understanding of how research has been cited.

The webinar offers a special coupon (site_webinar_september) for a 30% discount on individual plans.

Scite's browser extension is available for multiple platforms, showing citation information for research articles online.

Scite uses a deep learning model to classify citation statements into supporting, mentioning, or contrasting evidence.

Over 900 million citation statements have been analyzed to provide different citation types.

Scite partners with publishers to access content not freely available, enhancing its database of citation information.

The citation context provided by Scite helps interpret how research is used and understood in the academic community.

Scite's search function allows users to find the most supported research on a specific topic, offering a new dimension to traditional searches.

Users can create custom dashboards based on topics of interest, enhancing the ability to track and understand specific areas of research.

Scite's Reference Check tool helps authors, editors, and peer reviewers analyze the references of a manuscript for editorial concerns.

Scite provides data on editorial notices such as retractions and errata, helping to ensure research integrity.

The Scite Index offers a metric for understanding the overall support or contrast of citations received by journals.

Scite's new citation statement search allows users to search the entire database for expert analyses and opinions on virtually any topic.

Scite acknowledges the foundational work of Eugene Garfield, the creator of Web of Science, who discussed similar ideas in 1964.

Scite is continuously improving and values user feedback for further development and refinement of its services.

The webinar concludes with an invitation for questions and feedback, emphasizing Scite's commitment to user engagement.