Skip to Main Content

Learning goals

By the end of today's session you will:

  1. Be able to describe key elements of what is primary research in computer science and why it is important [take me there]
  2. Understand than there is a particular approach to reading primary research and be able to identify key elements of that approach [take me there]
  3. Be able to do some basic evaluation of a primary research article for quality and impact [take me there]
  4. Know key sources for primary research and the value of searching in multiple places [take me there]
  5. Know where to locate tools to help you manage your references [take me there]
  6. Know how to get more research help [take me there]

1. Primary Research in Computer Science

What is primary research in computer science? Primary research in computer science involves the process of conducting original research to create new knowledge or understanding. The approach uses a systematic method of investigation relating to a problem or question and using methods that will generate new data or insights.  

How is research data gathered? Methods of data collection may involve:

  • Experimental: Testing and evaluating algorithms, software, or hardware under controlled conditions.
  • Empirical: Gathering data through observations, surveys, or user studies.
  • Simulation: Using computer models to simulate complex systems or behaviors.
  • Theoretical: Developing and proving theorems or models

How is it similar to primary research in other disciplines? It involves the typical research steps of problem identification, literature review, hypothesis testing, and dissemination.

What are some unique characteristics? It has a computational focus, fast pace, more interdisciplinarity and collaborative, high diversity of evaluation methods, and the significant role of conferences.

Research in machine learning is wide ranging and rapidly evolving, here are some examples of topics under investigation:

  • Supervised learning techniques: Example: An automated medical diagnostic system to help detect disease from images
  • Unsupervised learning and clustering: Example: improving understanding customer behaviors and preferences
  • Reinforcement learning: Example: Developing autonomous driving technology
  • Natural language processing (NLP): Example: Real time language translation
  • Ethical AI and bias mitigation: Example: Reducing racial bias in facial recognition technology
  • Time series analysis and forecasting: Example: A forecasting model to predict renewable energy outputs


2. An Efficient Approach to Reading Primary Research

The Three Pass Method for Reading Scientific Articles

Read the paper in three passes not from beginning to end. Each pass builds on the previous.

  • First pass = quick scan.
    • Goal: get the big picture or bird's eye view of the paper
    • Don't take notes. Scan for titles, headings, date, what type of research is, what are the major conclusions
  • Second pass = re-read
    • Goal = grasp paper's content, but not the details. 
  • Third pass = interpret
    • Goal: understand the paper in depth.

Sample article 

Can AI help me understand a paper? Maybe! Tools are evolving all the time.
  • Try ChatPDF which allows you to "chat" with your PDF article using AI (for example you can ask ChatPDF to summaries the PDF or ask it what the methods or main conclusions of the article were).  
  • PDFgear is a pdf reader that integrates with chat-pdf  [video]

3. Evaluating Primary Research

There are many different elements to consider when evaluating primary literature. How you approach your evaluation will depend on the purpose of that evaluation - such as if you want to replicate the research or use it to inform your work or evaluate the impact a particular research article on the field.
Evaluating the quality and impact of a journal article
  • Research problem and relevance: Ensure the problem is clearly stated and understand its significance. Check for a comprehensive literature review.
  • Methodology assessment: Confirm the methods are appropriate and detailed enough for reproducibility, including data processing and algorithm configurations.
  • Data quality evaluation: Look at the data source, quality, and preprocessing steps. The data should be well-suited and properly prepared for the study.
  • Results and interpretation: Review the statistical analysis and practical applicability of the results. The discussion should address implications, limitations, and future work.
  • Peer review and citations: Check if the article is peer-reviewed and note its citations to gauge its impact and relevance in the field.
  • Evaluating a dataset that supports primary research. If there are problems with an underlying dataset then there may also be errors or issues with the outcomes of any research based on the data. [Tips from UOttawa]
  • Example of a retracted article
Is it published in a predatory journal? 

What is a predatory journal? These journals take advantage of authors by asking them to publish for a fee and then they do not provide peer-review or editing. They often offer very quick turnaround for publishing. Higher quality journals can take longer to publish because of the review processes. 

How to spot a predatory journal:

There are no foolproof ways to determine if a journal is predatory, but there is some detective work you can undertake if you want to investigation.

  • Look for clues on the journal or publisher homepage such as whether the journal is clear about their peer review process, if the editor or publishers are easy to contact and if its clear what fees will be charged. 
  • Check to see if your journal is listed on the Directory of Open Access Journals. The DOAJ carefully vets all journals for inclusion
  • Search online for the journal's name along with the term "predatory" to discover any relevant news stories or reports about the journal's potentially unethical practices.

4. Finding Primary Research

Key sources of computer science literature

6. Getting Help

Stop by the Library information desk or use chat for quick ando/or general help.

Connect with Francine via email ( or book an in person or virtual meeting using the Schedule Appointment button. 


Computer Science Librarian

Profile Photo
Francine May

Associate Dean, Collections and Research / Associate Professor, Library

Phone: 403-440-6128

SPoT - student perception of teaching