scite Webinar (June 15, 2021)

scite
15 Jun 202150:11

TLDRThe scite webinar, led by co-founder and CEO Josh Nicholson, introduces scite, a tool enhancing the understanding and reliability of scientific research through a new generation of citations. It classifies citations as supporting, contrasting, or mentioning, offering a more nuanced view of research impact and reproducibility. The tool is designed to help researchers, students, and professionals interpret scientific literature better by providing context to how and why studies are cited.

Takeaways

  • πŸ˜€ Josh Nicholson, the co-founder and CEO of Scite, introduced the platform with a focus on improving research through a new generation of citations.
  • πŸ”Ž Scite aims to provide more nuanced insights into how scientific studies are cited, moving beyond just citation counts to understanding the context and support within citations.
  • 🌐 Nicholson highlighted the challenges of scientific publishing, including the time-consuming and costly process, and the importance of publications in informing policy and future research.
  • πŸ“‰ The webinar discussed the reproducibility crisis in research, where many studies struggle to be replicated, affecting the reliability of scientific findings.
  • πŸ“ˆ Scite uses a browser extension and deep learning models to classify citations as supporting, contrasting, or mentioning, offering a more detailed view of a paper's impact.
  • πŸ” Scite allows users to search and filter citations based on various criteria, such as article type, author, and date, to find relevant research more efficiently.
  • πŸ“Š The platform offers visualization tools to represent citation networks, helping users understand the connections between different studies and their impact on the field.
  • πŸ“ Scite is working on adding more features, such as showing how authors cite others and providing more detailed information about individual citation statements.
  • πŸ“š The platform has seen significant growth, with around 32,000 users, and is particularly popular among students for its ability to aid in literature review and interpretation.
  • 🌟 Scite's information is being integrated into various journals and university repositories, expanding its reach and accessibility to researchers.
  • πŸ”— The company offers an API for bibliometric analyses and is open to partnerships that can further integrate Scite's citation insights into existing research tools and services.

Q & A

  • What is the primary goal of scite?

    -The primary goal of scite is to improve the way scientific research is understood and interpreted by focusing on the quality of citations, providing more nuanced information about how a paper has been cited, rather than just the quantity.

  • How does scite classify citations?

    -Scite uses a deep learning model to classify citations into supporting, contrasting, or mentioning. This classification helps users understand the context and relevance of each citation.

  • What is the significance of the 'supporting' classification in scite?

    -A 'supporting' classification indicates that the citing paper provides evidence or analysis that agrees with the cited paper, suggesting that the findings of the cited paper are robust and reliable.

  • Can scite's deep learning model classify citations perfectly?

    -No, the deep learning model is not perfect and may misclassify some citations. Users have the option to flag misclassified citations, which are then reviewed by the scite team.

  • How does scite help users stay updated with new research in their field?

    -Scite allows users to set up email alerts for new citations or publications that match their search criteria, helping them stay current with the latest research.

  • What is the purpose of scite's browser extension?

    -The browser extension for Chrome makes it easy for users to passively and easily use scite while reading scientific literature online, providing citation context and classification directly in the reference sections of articles.

  • How does scite address the reproducibility crisis in scientific research?

    -Scite addresses the reproducibility crisis by surfacing expert analyses and data that help users better understand and interpret research findings, making it easier to assess the reliability of studies.

  • What is the 'R factor' mentioned in the script?

    -The 'R factor' was initially an idea for a single score around a publication to indicate its quality. However, due to pushback from researchers, scite has shifted away from using a single score for individual articles or researchers.

  • How can scite be used to understand the work of a specific researcher or lab?

    -Scite can be used to identify the most cited, supported, or mentioned articles by a specific researcher or lab, providing insights into their impact and influence within their field.

  • What are some limitations scite faces in expanding its services to different disciplines like computer science?

    -One of the main limitations scite faces in expanding to disciplines like computer science is access to full-text articles from publishers such as IEEE and ACM, which do not have indexing agreements with scite.

Outlines

00:00

πŸ“š Introduction to SITE and Research Challenges

Josh Nicholson, the co-founder and CEO of SITE, introduces himself and the platform's mission to enhance research quality through a nuanced citation approach. He acknowledges the difficulties in scientific publishing, such as the high costs and the time-consuming nature of peer review, while emphasizing the importance of new findings in various fields. Josh highlights the challenges of staying updated with the overwhelming amount of published research and the issue of reproducibility in scientific studies, which SITE aims to address by providing detailed citation analysis.

05:01

πŸ” The Problem with Traditional Citation Metrics

Josh discusses the limitations of traditional citation metrics, which often lack the nuance needed to properly evaluate the impact and reliability of research. He shares his personal experience during his PhD, where the pressure to publish in high-impact journals led to the exclusion of contradictory data. SITE's goal is to move beyond mere citation counts and provide a more detailed analysis of how research is cited, including whether citations are supportive, contrasting, or simply mentioning the work.

10:03

πŸ› οΈ SITE's Solution: Deep Learning for Citation Context

The script explains how SITE uses a deep learning model to classify citation statements as supporting, contrasting, or mentioning. This classification helps users quickly understand the context of citations and evaluate the trustworthiness of research. The platform has been trained on a dataset of manually read citation statements to ensure accuracy, and users can provide feedback on misclassifications to improve the system.

15:04

πŸ“˜ The Importance of Citation Context in Research

Josh emphasizes the value of understanding the context in which research is cited, as it provides insights into the validity and impact of studies. He illustrates this with an example from his own research on chromosome segregation defects, showing how SITE can reveal the nature of citations and help users find related work and expert analyses.

20:04

πŸ”„ The Evolution of SITE's Reproducibility Metric

The script outlines the evolution of SITE's approach from focusing solely on reproducibility to a broader aim of surfacing expert analyses and data. Josh explains that while reproducibility is crucial, the platform now seeks to provide a more comprehensive understanding of research by making citation contexts easily accessible and interpretable.

25:06

🌐 SITE's Browser Extension and Citation Networks

Josh introduces SITE's browser extension for Chrome, which allows users to access citation information while reading scientific literature online. The extension enables users to see citation contexts directly in reference sections and provides a visual representation of citation networks, helping users understand the relationships between different studies.

30:06

πŸ“Š Analyzing Citations in Aggregate and Custom Dashboards

The script describes how SITE allows users to analyze citation information in aggregate, creating custom dashboards to explore topics, journals, universities, and funding sources. This feature enables users to identify trends, compare fields, and monitor the impact of research over time, providing a deeper understanding of the scientific landscape.

35:06

πŸ€– The Future of Citation Analysis with SITE

Josh concludes the presentation by discussing the future of citation analysis with SITE. He mentions the platform's plans to expand its analysis capabilities, including the addition of author information, reference manager plugins, and a reference check feature. He also highlights the importance of user feedback and the platform's commitment to continuous improvement.

40:10

🌟 Growing Adoption and Integration of SITE

In the final paragraph, Josh reflects on the growing adoption of SITE, with a particular increase in student users. He discusses the integration of SITE's information on the version of record for numerous publications and the strategy to embed SITE's citation analysis in various existing tools and services, aiming to make the platform an essential part of the research community's workflow.

Mindmap

Keywords

πŸ’‘scite

Scite is a platform designed to enhance the way research is evaluated and interpreted by focusing on the quality and context of citations. In the video, Josh Nicholson, the co-founder and CEO of scite, explains that the platform aims to move beyond mere citation counts and provide a more nuanced understanding of how research is used and perceived in the scientific community. For example, scite classifies citations as supporting, contrasting, or merely mentioning, allowing users to discern the nature of the relationship between cited and citing works.

πŸ’‘citations

Citations are references to previously published work within an academic or scientific paper. They are a fundamental part of scholarly communication, indicating the intellectual debt to other researchers' work. In the context of the video, citations are not just counted but are also evaluated for their supportive, contrasting, or neutral nature towards the cited work, as per scite's methodology. This approach helps in understanding the impact and reception of research findings within the scientific discourse.

πŸ’‘reproducibility crisis

The reproducibility crisis refers to the challenges faced in scientific research where the results of certain studies cannot be replicated by other researchers. In the video, Josh Nicholson discusses scite's role in addressing this issue by surfacing expert analyses and data that can help users better understand the reliability of research. The script mentions that the inability to reproduce study outcomes can cast doubt on the robustness of findings.

πŸ’‘deep learning model

A deep learning model is a type of artificial intelligence algorithm that is capable of learning and making decisions based on large amounts of data. In the video, scite utilizes a deep learning model to classify citation statements as supporting, contrasting, or mentioning. This model has been trained on a dataset of manually classified citation statements to enable the platform to analyze and categorize citations at scale.

πŸ’‘browser extension

A browser extension is a software component that adds functionality to a web browser. In the context of the video, scite has developed a browser extension for Chrome that allows users to easily access scite's citation analysis features while reading scientific literature online. This extension enhances the user experience by providing direct access to citation context and classifications without needing to navigate away from the article being read.

πŸ’‘editorial notices

Editorial notices are statements or annotations added to a published article that highlight corrections, retractions, or other amendments to the original publication. In the video, scite includes information about editorial notices in its analysis, helping users to be aware of any post-publication changes or concerns related to a research article.

πŸ’‘site index

The site index, as mentioned in the video, is a metric used by scite to provide a general overview of the support or contrast surrounding a body of research over time. It reflects the balance between supporting and contrasting citations, offering insights into the evolving consensus or debate within a scientific field. This index helps users to gauge the stability and reliability of research findings.

πŸ’‘preprints

Preprints are versions of scientific papers that have been shared before they undergo formal peer review. They are an increasingly popular way to disseminate research findings quickly. In the video, scite's inclusion of preprints in its analysis reflects the platform's commitment to staying current with the latest research, including works that have not yet been formally published or peer-reviewed.

πŸ’‘reference manager

A reference manager is software that helps scholars organize and manage their bibliographic references. In the video, scite is compatible with popular reference managers like Zotero and Mendeley, allowing users to integrate scite's citation analysis into their existing workflows. This feature enhances the utility of scite for researchers who rely on these tools for literature review and citation management.

πŸ’‘API

API stands for Application Programming Interface, which is a set of rules and protocols that allows different software applications to communicate with each other. In the video, scite offers an API for researchers to use for bibliometric analyses, enabling them to leverage scite's data and analysis capabilities in their own research projects or to create custom applications.

Highlights

Introduction to Sciite, a tool aimed at enhancing research through a new generation of citations.

Josh Nicholson, Co-founder and CEO of Sciite, shares his background in cell biology and the motivation behind Sciite.

The problem of scientific publishing's reliance on proxies of quality such as where it's published and who the authors are.

Sciite's focus on citations to provide more nuanced understanding of studies and their impact.

The challenge of scientific publishing's costs and the reproducibility crisis in research.

How Sciite surfaces expert analyses and data to improve the reliability and interpretation of scientific research.

Personal anecdotes from Josh Nicholson's PhD experience highlighting the pressures of publishing in competitive journals.

The idea behind Sciite to introduce a metric around reproducibility and improve scientific research's reliability.

Demonstration of how Sciite works, showing the world of scientific literature with and without Sciite's insights.

The introduction of Sciite's browser extension for Chrome that enhances the reading of scientific literature online.

How Sciite provides citation context, making it easier to understand and interpret studies.

The explanation of the difference between supporting, contrasting, and mentioning in Sciite's citation classification.

The importance of not relying solely on numbers but understanding the nuance behind each citation.

The ability of Sciite users to flag misclassified citations, contributing to the accuracy of the system.

Sciite's visualization reports that show citation networks and help in understanding the field of work.

How Sciite can be used to analyze groups of articles, making it easier to understand a topic area or track research.

The introduction of Sciite's dashboard feature for creating custom reports on topics, journals, universities, and funders.

The discussion on the limitations of Sciite in certain fields like computer science due to access to full texts.

The future plans for Sciite, including partnerships with publishers and embedding information in university repositories.

The availability of Sciite's API for research purposes and the potential for bibliometric analyses.

Josh Nicholson's invitation for feedback and the emphasis on continuous improvement of Sciite based on user input.