Using AI-powered Tools for Literature Reviews

Scite is a tool which offers a quantitative and qualitative insight into how publications cite each other. It has 1.2 billion citation statements extracted and analysed from 181 million articles, book chapters, preprints and datasets. It accesses full text articles and uses a deep learning model to tell you

• How many times an item was cited by others.
• How it was cited by others by displaying the surrounding text within that paper.
• Whether each citation offers supporting or contrasting evidence, or simply mentions it.
• See an example publication.

Scite users can

• Search citation statements: discover and understand whether an article provides supporting or contrasting evidence without reading the full article. 
• Customise search criteria, including selecting trusted journals only and setting up year range and publication types.
• Create your own dashboard: evaluate groups of articles in a single place and receive notifications when new supporting or contrasting citations are made.
• Visit Journals or Institutions' dashboards: compare rankings or find their publications.
• Visualise research: discover and evaluate how articles cite each other through interactive visualisations.

More about Scite

Elicit is a free research tool using large language models to automate parts of researchers’ workflows. It is especially useful for empirical research like randomised controlled trials.  After entering a research question, Elicit will display relevant papers and summaries of key information about those papers in an easy-to-use table. The steps are

1. Search over 200 million papers from Semantic Scholar database using natural language. You can then select the 400 closest papers by saving titles and abstracts using paraphrase-mpnet-base-v2.
2. Re-rank those 400 papers using GPT-3 Babbage model, and further filter to the top 20 using castorini/monot5-base-msmarco-10k. Elicit then allows you to select the top 8 papers.
3. Find and parse PDFs of open access papers using Unpaywall and Grobid.
4. Summarise the top papers and provide short answers to your search question.
5. For each of the top papers, extract key information like outcomes measured, interventions and sample size, and display them in columns. Users can download a CSV or .bib file to import into reference managers like EndNote, Mendeley and Zotero.

More about Elicit

SciSpace/Typeset is an AI-powered research platform with tools to discover, explore, understand, and write research, which includes below key features:

• Chat with PDF: summarize and answer questions based on the content from multiple PDF documents.
• Literature Review: find related papers by uploading seed papers, or searching keywords in natural language.
• AI Writer: use AI suggestions to expand notes to paragraphs.
• Find Concepts: discover concepts from 285M research papers.
• Paraphraser: increase fluency of writing.
• Citation Generator: generate citations in APA and other formats.
• Extract Data: get summary, methods, conclusions and findings from multiple PDFs, up to 50 parameters can be customised in table fomat.
• AI Detector: analyze essays and research papers for AI content.
• Convert scientific PDFs into video abstracts and presentations.

More about SciSpace

Consensus is a search engine powered by AI in scientific research. It uses Semantic Scholar data and synthesize both topic-level and paper-level insights.

Key Features of Consensus:

• Consensus Meter 

Consensus uses OpenAI’s GPT-4 model to generate a one-sentence summary of the top 10 studies. For yes/no questions, a custom model classifies results as “yes,” “no,” or “possibly,” shown in the Consensus Meter, which displays a distribution of supporting and contrasting evidence.

• Advanced Search Filters
  • By study design
  • By sample size
  • By methodology
  • By SJR quartiles

More about Consensus

Connected Papers is a visual tool to help researchers find and explore papers relevant to their field of work.

• It analyses papers from Semantic Scholar to create a graph, and then selects the most relevant papers.
• In the graph, papers are arranged according to their similarity. That means even papers that do not directly cite each other can be strongly connected and very closely positioned. 
• The similarity metric is based on the concepts of co-citations and bibliographic coupling. According to this measure, two papers that have highly overlapping citations and references are presumed to have a higher chance of addressing a similar subject.
• The algorithm then builds a Force Directed graph to distribute the papers in a way that visually clusters similar papers together and pushes less similar papers away from each other. Upon node selection it highlights the shortest path from each node to the original paper.

More about Connected Papers

Inciteful has developed two tools to date

• Paper Discovery tool builds a network of papers from one seed paper. It uses network analysis algorithms to find similar papers, important papers, as well as prolific authors and institutions. 
• Literature Connector tool helps interdisciplinary scholars bridge two domains. It allows users to enter two papers and it will generate an interactive visualisation map showing you how they are connected by the literature.

More about Inciteful

Litmaps generates an interactive map of relevant articles that relate to a seed paper. The most recent articles appear on the right hand side of the map, the most cited articles appear at the top and the lines show the citations in-between. Users can click on each dot and add more concepts to expand the map.

More about Litmaps
‍