Ai2 Paper Finder works in a semi-rigid flow: a predefined structure that is influenced at key points by various LLM decisions. First, the query arrives at the query analyzer, a component that looks at the query and breaks it down to intents and components. The query planner takes the analyzed query, and comes up with an execution strategy. This strategy depends on the search goals, and also on what's available in the different APIs and what will be efficient to execute. Our current implementation routes to one of several pre-defined sub-flows: specific paper search, semantic search with potential metadata constraints, pure-metadata queries, and queries that involve an author name. Find more details about the tool. [information updated on 2025.05.26]

Ai2 ScholarQA follows a RAG-based, multi-step prompting workflow using Claude Sonnet 3.5 model. It relies on a corpus of open-access papers (from Semantic Scholar, arXiv, and other free sources). Ai2 ScholarQA is meant to satisfy literature searches that require insights from multiple relevant documents, and synthesize those insights into a comprehensive report. After receiving a query, the system first queries the index for the top k passages. These passages are further re-ranked with a pretrained transformer model and the top 50 candidates are retained for further processing. The answer generation is a 3-step process driven by prompts to an LLM: (1) quote extraction, (2) answer outline and clustering, (3) report generation. Find more details about the tool. [information updated on 2025.03.07]

Consensus uses language models (LLMs) and purpose-built search technology (Vector search) to surface relevant papers. The current source material used comes from the Semantic Scholar. Consensus run a custom, fine-tuned language model over the entire corpus of research papers and extract the Key Takeaway from every paper. Next, elements such as ‘what’, ‘is’ and ‘are’ are removed from the query and a combination of keyword search and Vector search is performed on the abstract and title of all articles. Find more details about the tool. [information updated on 2025.03.07]

Elicit searches across over 126 million academic papers from the Semantic Scholar corpus across all academic disciplines, including preprints (from arXiv, bioRxiv, and other repositories). It also includes non-open-access papers. Elicit answer will be based on the full text of the paper if it is available via open access, otherwise Elicit will answer based on just the abstract. Find more details about the tool. [information updated on 2025.03.07]

ORKG Ask employs a sophisticated pipeline to ensure you receive the most relevant information from the vast CORE dataset of research articles. The system uses an embedding model to encode your query and searches for semantically similar research articles using a vector store with an Approximate Nearest Neighbor (ANN) algorithm. ORKG Ask leverages the following advanced technologies to deliver its robust search and extraction capabilities: Qdrant vector store, Nomic embedding model and Mistral Instruct 7B v0.2 large language model. Find more details about the tool. [information updated on 2025.03.07]

Perplexity uses cutting-edge language models like GPT-4 Omni and Claude 3 to understand the queries context and nuances. It searches the internet in real time, gathering insights from diverse sources. Four ways to search: Auto (adapts to the question), Or (more sources and detailed answers), Reasoning (advanced problem-solving) and Deep Research (in-depth report on complex topics). Find more details about the tool. [information updated on 2025.03.10]

SciSpace sources research articles and related data from multiple platforms including Semantic Scholar, OpenAlex, Google Scholar, and Microsoft Academic Graph. It uses GPT 3.5 for basic plans and GPT 4o for other plans. Scispace's search engine uses a combination of semantic and text-based methods, focusing on titles and abstracts. Citation analysis is not used in the ranking of selected articles. Find more details about the tool. [information updated on 2025.02.27]

Undermind covers all fields of science using the Semantic Scholar database. It employs a blend of lexical or keyword search and embedding-based vector or semantic search. Instead of a quick search, Undermind does a more thorough search in several steps that can take up to 8 minutes. It uses a search algorithm that, after finding relevant content, adapts to conduct successive keyword and citation searches. Additionally, it directly employs large language models like GPT-4 to classify papers into three levels of relevancy. Find more details about the tool. [information updated on 2024.12.11]