Citrus Search

Description

Citrus Search provides a novel approach to navigating scientific literature, moving beyond traditional text queries that often result in numerous false positives. By leveraging similarity-based search, researchers can gain a comprehensive overview of their field with a single query. The core functionality involves selecting "seed papers" – foundational or highly relevant articles – and then exploring publications that are closely related.

What constitutes "related" is determined by user-selectable similarity measures. The "Citation Network" measure identifies papers with similar connections within the citation graph, revealing influential works and their connections. The "Content" measure, on the other hand, finds papers with similar concepts, ideas, and research questions by analyzing their titles and abstracts. This is powered by sophisticated graph and text-based machine learning techniques that compute the similarity between papers.

Citrus Search indexes a vast dataset provided by Semantic Scholar's Open Research Corpus, which encompasses over 200 million publications and approximately 2 billion citations. This extensive coverage ensures that users can discover a wide range of relevant research. The platform presents key contributions at a glance, often visualized on a timeline, which is crucial for understanding the evolution of a research area and ensuring no significant papers are overlooked. This is particularly beneficial for identifying work that might use different terminology or taxonomies than standard text searches would capture.

The service is under active development, with a commitment to continuous improvement based on user feedback. Researchers experiencing bugs or having suggestions for enhancements are encouraged to use the provided feedback form or contact the development team directly via email. Citrus Search aims to be an indispensable tool for academics, students, and R&D professionals seeking to efficiently and effectively explore complex research landscapes.

Citrus Search's Core Features

• Similarity-based search for scientific literature

• Select seed papers to initiate search

• Explore closely related publications

• Citation network similarity measure

• Content-based similarity measure (title and abstract)

• Graph and text-based machine learning techniques

• Overview of research fields

• Timeline visualization of important contributions

• Indexes Semantic Scholar's Open Research Corpus

• Covers over 200 million publications

• Includes around 2 billion citations

• Helps avoid missing relevant papers

• Addresses limitations of traditional text queries

How to use Citrus Search?

1. 1

   Select Seed Papers: Choose foundational or highly relevant articles to start your search.

2. 2

   Choose Similarity Measure: Decide whether to prioritize citation network connections or content similarity.

3. 3

   Explore Related Work: Review the list of publications identified as closely related to your seed papers.

4. 4

   Gain Research Overview: Utilize the timeline view to understand the landscape of important contributions.

5. 5

   Refine Search: Adjust seed papers or similarity measures for more targeted results.

6. 6

   Provide Feedback: Share insights or report issues to aid in active development.

Citrus Search's Use Cases

• Literature Review
• Research Discovery
• Identify Key Contributions
• Avoid Missing Papers
• Explore Research Trends
• Seed Paper Analysis

FAQ from Citrus Search

• What is Citrus Search and how does it work?

  Citrus Search is a similarity-based engine for scientific literature. It allows users to select seed papers and find closely related publications using citation network or content similarity measures, powered by machine learning.

• How does Citrus Search help researchers find relevant articles?

  It helps by going beyond traditional text searches, which can miss papers or yield many false positives. Citrus Search identifies related work based on connections in citation networks or conceptual similarities in content.

• What kind of similarity measures does Citrus Search offer?

  Citrus Search offers two primary similarity measures: 'Citation Network' to find papers with similar connections in the citation graph, and 'Content' to find papers with similar concepts in their abstract and title.

• What data sources does Citrus Search utilize?

  Citrus Search indexes data provided by Semantic Scholar's Open Research Corpus, which includes over 200 million publications and around 2 billion citations.

• Can Citrus Search help me avoid missing important research papers?

  Yes, by identifying papers that might use different terminology or taxonomies than standard text searches, and by providing an overview of contributions, it helps ensure you don't miss key publications.

• How is the similarity between papers computed in Citrus Search?

  The similarity is computed using graph and text-based machine learning techniques behind the scenes.

• Is Citrus Search still under development?

  Yes, the site is under active development, and user feedback is encouraged for improvements and bug reporting.

• Who is the target audience for Citrus Search?

  Citrus Search is designed for researchers, academics, students, and anyone needing to efficiently explore and understand scientific literature in a specific field.

• What is the main benefit of using Citrus Search over traditional search engines?

  The main benefit is its ability to provide a more comprehensive and accurate overview of relevant research by focusing on similarity rather than just keyword matching, reducing false positives and missed papers.

• Can I see an overview of my research field with Citrus Search?

  Yes, Citrus Search provides an overview of relevant articles in a research field, often presented on a timeline to highlight important contributions.