Finding scientific literature - PeerWriting.org

4 levels

From our personal experience, there are currently 4 levels of tools. We categorize them based on which part of the paper (or its metadata) that is searched, and the output they are able to provide: ranking, relation, data extraction

Rank papers, based on word matching in title & keywords— Cible+, Scopus, PubMed, Google Scholar, Science Direct, Web of Science, etc

Map network of citations— Connected Papers, Rabbit Research, Open Knowledge Map

Provides summary, influence-in-literature metrics, and rank papers, based on abstract and paper — Semantic Scholar

Extract & creates a table of information based on the AI-reading of the abstract and paper — Elicit

Note that all methods use the title, keywords, and citation count. In most cases, we can’t explain the methods of each toos, either because they are not disclosed, (eg), or is too complex and beyond our scope.

Conventional search engine

Research Rabbit

visit Research Rabbit

Papers	>200M+ (it accesses semantic scholar database)
Creators	independents, based on Microsoft Academic Graphs
Specificity	Mapping of paper relations based on citation
Fields	all
Main Features	1. Maps all citing paper based on 1 or more paper seed
	2. Promotes literature navigation via the visual interface. Allows traceback to the original query
	3. Exploration of Authors and co-authors
	4. Paper feed based on papers contained in your folders.
	5. Zotero integration
Best for	All types of search (general, or specific)
	Exploring the research landscape of a topic
	Litterature monitoring
Alternative	Connected Papers
	OpenKnowledge Maps
	Litmap
	Inciteful
	Microsoft academic graphs

Rabbit research and alternatives offer a mapping of a paper’s landscape, based on citing literature

OpenKnowledge Maps creates groups of concepts based on the top 100 results, of keyword search

Semantic Scholar

visit Semantic Scholar General demo

Papers	200M+
	from 30+ publishers
Creators	Allen Institute for AI
Specificity	Natural Language Processing AI (GPT3-like)
Fields	all
Main Features	1. AI-Assisted Ranking → Method for ranking based on a larger parameter set
	2. “TLDR”, an extreme summarization of papers [source], only if the entire paper is available open-source
	3. Influence in citing literature→ AI estimate of the importance of a paper in the literature citing it: ranks by most influenced
	4. Citation intent → AI estimates the reason for citing a paper based on category: background, method, results.
	5. Paper feed based on papers contained in your folders. Info
	6. Semantic reader (beta)→ AI contextual definition of abbreviation and variables. Info
Best for	All types of search (general, or specific)
	Litterature monitoring
	Weighing the influence of paper in a field

An AI ranks papers based on a series of parameters extracted from the paper metadata and the abstract (source)

An NLP-AI extracts a TDLR (20 words long)

“An alternative to abstracts, TLDRs of scientific papers leave out nonessential background or methodological details and capture the key important aspects of the paper, such as its main contributions”. [source]

Elicit

visit Elicit

Papers	>200M+ (it accesses semantic scholar database)
Creators	ought.org
Specificity	AI-based information extraction and summary from abstracts
Fields	all
Main Features	1. AI (GTP-3) interprets research questions expressed in natural language
	2. AI extracts data and information from papers
	3. “Take-out from the study” are generated
	4. Information to be extracted is user-designed
	5. AI brainstorms research question
	6. provide possible critiques of paper based on citations in other papers
Best for	All types of search (general, or specific)
	Quick and summarized response to a research question
	A rapid overview, great for non-expert topic