Ai2 Paper Finder: LLM-Based Academic Paper Search System 147

What is ai2 Paper Finder?

Ai2 Paper Finder is a literature search system released by the Allen Institute for Artificial Intelligence on March 28, 2025. It is based on a large language model and uses an iterative search approach that mimics the human process of finding literature, helping users discover niche, hard-to-find papers. Users enter the topic they want to search, the AI automatically finds all relevant papers, provides a relevance summary for each paper, and supports filtering by relevance, time, and other criteria. Paper sources include arxiv and other institutions, with a corpus of over 8 million full-text papers and more than 108 million abstracts.

Ai2 Paper Finder is currently free and requires no registration, making it easy to use. It is suitable for researchers, scholars, students, and other groups.

Core Features of Ai2 Paper Finder

• Iterative Search: Mimics the multi-step process by which human researchers find literature, including decomposing queries, tracking citations, assessing relevance, and follow-up queries. Users can directly enter query requirements described in natural language, and the system will automatically analyze and perform iterative searches.

• Deep Coverage: Aimed at discovering niche, hard-to-find long-tail papers, rather than just popular results.

• Intelligent Summaries: Provides a brief summary of each paper's relevance to the query.

Ai2 Paper Finder Search Process

• Specific Literature Search: Searches in parallel through three strategies, including using the Semantic Scholar title search API, searching sentences containing core keywords to find referenced literature, and asking the LLM for paper titles and verifying their existence.

• Semantic Search: Uses an LLM to create multiple paraphrased versions of semantic criteria, and searches each separately in three different dense indexes, with each index returning paper snippets of varying lengths. It also adds candidate literature based on citations in returned snippets, and uses an LLM to assess the relevance of candidate papers.

Advantages of Ai2 Paper Finder

• Ease of Use: Users do not need to simplify queries into keywords; they can directly enter a natural language description, and the system will automatically analyze and retrieve. For a single piece of literature, it provides relevance reasons, citation counts, and relevant discussions in the body. Clicking "show evidence" allows quick viewing of relevant discussions in the body.

• Cost-Effectiveness: Currently free, with no registration required.

• Result Presentation: The right side allows further filtering by year, journal, and author, and filtered literature can be exported with one click.

Applications of Ai2 Paper Finder

• Suitable for researchers, scholars, students, and other groups who need to conduct in-depth literature research.

• Academic Research: In scientific research projects, it helps researchers quickly find frontier literature and classic papers in related fields, providing theoretical support and reference for research topics.

• Paper Writing: Provides literature retrieval services for authors writing academic papers, helping them find cited literature, enrich paper content, and enhance academic value.

• Literature Review: When conducting a literature review, it can help researchers comprehensively and systematically collect literature on a specific topic and sort out the research context.

• Course Learning: When completing coursework assignments, graduation projects, or writing course papers, students can use it to quickly find the academic materials they need.

Usage Suggestions

• Enter a complete research question

  For example: "The relationship between social capital and political participation depends on cultural context and institutional trust factors, especially in emerging democracies." rather than a single keyword.

• Combine filtering features

  Use filtering conditions such as year and journal to narrow down the scope, for example focusing on papers from Nature sub-journals in the past 5 years.

• Use deep search mode

  When preliminary results are unsatisfactory, enter "work harder" to trigger an expanded search.

Related News:

60 Must-Have Academic Research Tools for Paper Writing for Researchers