Brief Description of Resource
Many students have reported difficulty finding peer-reviewed journal articles, as well as reading and accessing the scientific literature. This activity is meant to guide them through locating papers that are valid scientific sources using both traditional and modern AI means, and exploring AI tools to find related articles and help students better understand the papers that they find.
Instructions for Use/Implementation
The below worksheet, Modern Ways to Access Scientific Papers, can be printed or distributed digitally for students to work through as an in-class activity. It can be done individually or as a group
AI Tools Used
Consensus
Connected Papers
SciSpace
Explainpaper
Prep Time Needed
5-10 minutes for printing or posting to LMS
Classroom Time Needed
25-40min
Student Learning Outcomes
Know how to find reliable scientific sources online
Be able to evaluate peer-reviewed scientific journal articles
Assessment Considerations
Participation credit recommended
Recommended Course Size
Any size, can be group activity for large lecture course or individual/group in recitation, lab or co-seminar
Recommended Discipline(s)
Any science discipline, or applied to any other area which relies heavily on understanding peer-reviewed literature
Student Role(s)
Students work individually or in groups to navigate online resources, and answer questions about exploring and comparing the usefulness of different tools
Instructor Role
Instructor or TAs can assist with any technical issues, or facilitate with further probing questions on how to find and evaluate scientific papers. Feel free to email author for .docx version
Modern ways to access Scientific Papers
Scientific journal articles are much more than just a news article or informational website (WebMD, Mayo Clinic, etc. are not valid resources). What makes them unique is that they are typically peer-reviewed by other experts in their field to ensure a high standard and greater reliability.
The peer-review and journal editing process is very rigorous, and can take between 6 months to a year prior to being published, with several rounds of revisions.
This is why legitimate scientific papers are required to be the references in your poster and manuscript assignments. Many journal articles can be found on a journal website but also have a PDF version of the paper itself. If there is only a website available, that is likely not peer-reviewed, and should not be included as one of your references.
Part I – Kicking it Old-School
Historically, since the 1980’s-1990’s, online databases have replaced a trip to the library to read journal articles in their printed version (like a thick magazine or thin book).
Open Web of Science (www.webofscience.com) and type in a topic you are interested in researching (either for this class, or in general). On the left hand menu, find the “Document Types” section and select “Article” then hit “refine”.
- a) Look through the search results and write the title of one of the papers you found.
Hover over the Journal name (in red box), click on it to find more information.
b) What is the Impact Factor for the journal your article was published in (if any)? The impact factor is a measure of how influential the journal is (readership and citations).
Congrats, you found a peer-reviewed article! How do we know this? The Web of Science is a database which only has peer-reviewed articles in it. Easy!
Another database you can use is called Academic Search Premier or EBSCO, which you get to via CU Library at (https://research.ebsco.com), you may have to select your institution and either be on campus or login through CU Library page off-campus.
Here you can select “Peer-Reviewed Articles” to ensure your reference is trustworthy.
2. Navigate to the EBSCO search database (https://research.ebsco.com) and search using the same terms you used in the Web of Science database in Q1.
a) On the first page of results do you see the same scientific article that you found (and wrote the title for) in Question 1?
b) Why do you think that is?
3. Do you think it is easy to tell whether an article is peer-reviewed using EBSCO?
The final old-school way we’ll discuss that researchers use to find papers is Google Scholar (https://scholar.google.com). This is a great tool for those familiar with Google, but not efficient because this database allows non-peer-reviewed articles as well as those that are more reliable/peer-reviewed. So what do we do?
Navigate to Google Scholar (https://scholar.google.com) and type in a topic to search for, one that you might be able to use for your research project.
4. a) Select an interesting paper from your search results, and write down below which journal it is published in. Ask the TAs if you need help.
b) Do an internet search of the journal name. Most publishers have on their website an “About” page, which talks about the process of how authors can submit a journal and what is required to be published. Sometimes it says very explicitly that it is peer-reviewed, other times you need to find the page that says “information/instructions for authors”, “submitting an article” or something similar.
Navigate through their site, is it peer-reviewed?
5. Of the three old-school databases you have used so far (Web of Science, EBSCO, Google Scholar), which do you prefer for finding peer-reviewed scientific journal articles, and why?
6. Choose your favorite database and enter keywords related to your project. Skim the abstracts that you find, to choose a journal article related to your research that you might want to use as a reference in your poster or manuscript. Download the PDF, you will need a PDF for uploading for later in the worksheet.
Part II – Modern Methods
Now let’s fast forward to modern times. Artificial Intelligence (AI) has been used and developed by computer engineers and scientists for decades, but only became mainstream recently. Many AI tools have been developed in 2020-2023 for widespread, end user-friendly purposes, including for finding and reading scientific papers.
Finding scientific papers-
Consensus (https://consensus.app) uses large language models (LLMs) and machine learning to surface relevant papers. That means that the method in which the algorithm searches for papers according to your query/prompt is not only different than the old-school methods (and those algorithms are different from one another depending on the database being used), but is also more adaptive (uses modern AI to “learn”) to your search terms which can be more complex and conversational.
According to their site, Consensus only searches through peer-reviewed scientific research papers, to “find the most credible insights to your queries.” That’s GREAT news! That means you can use the papers from here in your poster or manuscript.
7. Navigate to Consensus (https://consensus.app).
a) First enter one of the search terms you tried earlier in one of the “old-school” databases. Write down the search term/keyword(s) that you will try first here-
b) Read through the search results. Depending on what you are seeing, you can expand or narrow your query to be along the lines of what you are interested in, beyond just using keywords (as you would have done for an old-school method, or general internet search).
For example, instead of just a keyword- enter a question (even a yes/no one), ask about the relationship between different concepts, or enter a conversational command for their generative AI copilot.
Try out three different more complex/conversational queries to vary your results, write down the prompts that you attempted here-
i.
ii.
iii.
c) What did you think of this new resource compared to searching the old-school databases in questions 1-6? Did you find it easier or more difficult to use?
d) Due to the way this AI resource works, you may end up finding a paper that you find more interesting (and want to read in more detail) than the one you downloaded earlier. If so, download the PDF to use for the rest of the worksheet.
8. Researchers conduct systematic literature reviews, where they can read hundreds of papers related to their topic. AI cannot replace a thorough review of the scientific literature, however it can assist in finding related papers to read and understand.
Navigate to ConnectedPapers (https://www.connectedpapers.com). Type in the title of the paper you like the most so far (or several of its keywords). Click on your paper from your results (or a different paper option you find interesting). Use the map in the center to find highly cited (a lot of connections) papers that are related to your topic. In general newer (darker colored) is better since it is more up-to-date with recent breakthroughs. Skim through the abstracts on the right hand window. You can download any paper you find interesting from the three dot “More” button on the top right.
a) Which paper from your search is the most highly cited (largest circle)? Write the first author last name and year:
Part III- Now what? Understanding scientific papers
Some journal articles can be dense and filled with jargon, especially for students, early career researchers and those changing career fields. This can make reading them a bit of a slog.
Aside from conducting internet searches on different ideas and topics, a few AI tools have been developed. These are a good starting point to help novices understand highly-technical, disciplinary specific articles, along with reading a lot of papers from the field more thoroughly.
Choose your own Adventure.
Choose one of the following AI programs to explore: SciSpace or Explainpaper
You will need to upload the paper that you want to understand further, as a PDF. You can choose the paper you downloaded earlier (old-school database method), or a paper you found more interesting with the modern AI search tools.
SciSpace explorers start here:
Navigate to SciSpace (https://typeset.io), in the top right corner hit “+Upload PDF” to upload your paper of interest into the program.
In the right-hand window at the bottom (immediately above the chat box), choose from the list of prompts, “Explain abstract of this paper in 2 lines”.
a) Do you find the summary to be overly complicated or easy to understand, compared to the actual abstract in the paper?
Scroll through the paper and find some sentences or a paragraph that is super confusing and muddy to you. Highlight just that area of text and click “explain text”.
b) Does the AI explanation make more sense to you than the original text did?
Hopefully now you have a good idea of how this tool works. If you are interested in using it further, it will ask you to sign up for an account at this point (use a personal email rather than a colorado.edu email).
c) If you sign up, select one of the general prompt suggestions, and report below if you think that the results of your query were useful.
- If you do not wish to sign up, write below which general prompt you would have chosen in order to get to know your paper better (what is most confusing to you).
d) How can you envision using this tool for your independent research project?
Explainpaper explorers start here:
Navigate to Explainpaper (explainpaper.com) and hit “Try for Free.” You will need to sign up first (use a personal email rather than a colorado.edu email). Hit the “Upload PDFs” button to upload your paper of interest into the program, and hit “Begin Reading”.
On the top of the right window, hit “paper” to see the “gist” or summary of the paper.
a) Do you find the summary to be overly complicated or easy to understand, compared to the actual abstract in the paper?
Scroll through the paper and find some sentences or a paragraph that is super confusing and muddy to you. Highlight just that area of text (it can be a little tricky/buggy to get the right section), and then on the right hand window under “Explain”, choose the level of ‘Undergrad’ and hit the “Explain” button.
b) Does the AI explanation make more sense to you than the original text did?
Underneath your “Explain” result, click on “Ask a follow-up” and ask one of the following:
- Summarize the findings.
- Why is this paper important?
- What is your critique of this paper?
- What does figure 1 tell us?
- What is some prerequisite knowledge for this paper?
c) Do you think that the results of your query were useful for getting a better idea of your paper?
d) How can you envision using this tool for your independent research project?