Research

Accelerate your dissertation literature review with AI

Dr. Stephen Enright-Ward
Accelerate your dissertation literature review with AI

Introduction

Dissertation writing is part of being a graduate student. There are many different ways to organise your research, and several steps to this process. Typically, the literature review is an early chapter in the dissertation, providing an overview of the field of study. It should summarise relevant research papers and other materials in your field, with specific references. To understand how to write a good literature review, we must first understand its purpose. The goals of a literature review are to place your dissertation topic in the context of existing work (this also allows you to acknowledge prior contributions, and avoid accusations of plagiarism), and to set you up to show you are making a new contribution to the field. Since literature review is repetitive, many students find it tedious. While there are some traditional tools and techniques to help, covered below, they tend to be cumbersome and keyword-based. For this reason, we built a better tool for research and literature review, which I describe in the last section. You can see the Lateral tool in action, and how it makes the literature review a lot easier. To sign up to the tool, click here.

1. Different kinds of reading

We can divide the activity of reading for research into three different kinds: 

  1. Exploratory reading, mostly done in the initial phase;
  2. Deep reading of highly informative sources; and 
  3. Broad, targeted skim reading of large collections of books and articles, in order to find specific kinds of information you already know exist.

1.1. Exploratory reading

Initially, a research student will need to read widely in a new field to gain fundamental understanding. In this early stage, the goal is to explore and digest the main ideas in existing research. Traditionally, this phase has been a manual process, but there is a new generation of digital tools to aid in getting a quick overview of your field, and more generally to organise your research. This stage can happen both before and after the research topic or question has been formulated. It is often unstructured and full of serendipitous (“happy accidental”) discovery  — the student’s job is to absorb what they find, rather than to conduct a targeted search for particular information.

Put another way: You don’t know what you’re looking for ahead of time. By the end of this phase, you should be able to sketch a rough map of your field of study.

1.2. Narrow, deep reading

After the exploratory reading phase, you will be able to prioritise the information you read. Now comes the second phase: Deep, reflective reading. In this phase, your focus will narrow to a small number of highly relevant sources — perhaps one or two books, or a handful of articles — which you will read carefully, with the goal of fully understanding important concepts. This is a deliberative style of reading, often accompanied by reflective pauses and significant note taking. If the goal in the first phase was sketching a map of the globe, the goal in this second phase is to decide which cities interest you most, and map them out in colour and detail.

1.3. Broad, targeted reading

You have now sketched a map of your field of study (exploratory reading), and filled in some parts of this map in more detail (narrow, deep reading). I will assume that by this point, you have found a thesis question or research topic, either on your own, or with the help of an advisor. This is often where the literature review begins in earnest. In order to coherently summarise the state of your field, you must review the literature once again, but this time in a more targeted way: You are searching for particular pieces of information that either illustrate existing work, or demonstrate a need for the new approach you will take in your dissertation. For example, 

  • You want to find all “methodology” sections in a group of academic articles, and filter for those that have certain key concepts;
  • You want to find all paragraphs that discuss product-market fit, inside a group of academic articles.

To return to the map analogy: This is like sketching in the important roads between your favourite cities — you are showing connections between the most important concepts in your field, through targeted information search.

Discover what matters with Lateral AI

The revolutionary AI solution that rapidly finds information across documents for you! Sign up Early access. No credit card required.

2. Drawbacks of broad targeted reading

The third phase — broad, targeted reading, where you know what kind of information you’re looking for and simply wish to scan a collection of articles or books to find it — is often the most mechanical and time consuming one. Since human brains tend to lose focus in the face of dull repetition, this is also a tedious and error-prone phase for many people. What if you miss something important because you’re on autopilot? Often, students end up speed- or skim reading through large volumes of information to complete the literature review as quickly as possible. With focus and training, this manual approach can be efficient and effective, but it can also mean reduced attention to detail and missed opportunities to discover relevant information. Only half paying attention during this phase can also lead to accidental plagiarism, otherwise known as cryptomnesia: Your brain subconsciously stores a distinctive idea or quote from the existing literature without consciously attributing it to its source reference. Afterwards, you end up falsely, but sincerely believing you created the idea independently, exposing yourself to plagiarism accusations.

3. Existing solutions to speed up literature reviews

Given the drawbacks of manual speed- or skim-reading in the broad reading phase, it’s natural to turn to computer-driven solutions. One popular option is to systematically create a list of search term keywords or key phrases, which can then be combined using boolean operators to broaden results. For example, in researching a study about teenage obesity, one might use the query:

  • “BMI” or “obesity” and “adolescents” and not “geriatric”,

to filter for obesity-related articles that do mention adolescents, but don’t mention older adults.

Constructing such lists can help surface many relevant articles, but there are some disadvantages to this strategy:

  1. These keyword queries are themselves fiddly and time-consuming to create.
  2. Often what you want to find is whole “chunks” of text — paragraphs or sections, for example — not just keywords.
  3. Even once you have finished creating your boolean keyword query list, how do you know you haven’t forgotten to include an important search query?

This last point reflects the fact that keyword searching is “fragile” and error-prone: You can miss results that would be relevant — this is known as getting “false negatives” — because your query uses words that are similar, but not identical to words appearing in one or more articles in the library database. For example, the query “sporting excellence” would not match with an article that mentioned only “high performance athletics”.

4. Lateral — a new solution

To make the process of finding specific information in big collections of documents quicker and easier — for example, in a literature review — search, we created the Lateral app, a new kind of AI-driven interface to help you organise, search through and save supporting quotes and information from collections of articles. Using techniques from natural language processing, it understands, out-of-the-box, not only that “sporting excellence” and “high-performance” athletics are very similar phrases, but also that two paragraphs discussing these topics in slightly different language are likely related. Moreover, it also learns to find specific blocks of information, given only a few examples. Want to find all “methodology” sections in a group of articles? Check. How about all paragraphs that mention pharmaceutical applications? We have you covered. If you’re interested, you can sign up today.

5. Final note — novel research alongside the literature review

Some students, to be more efficient, use the literature review process to collect data not just to summarise existing work, but also to support one or more novel theses contained in their research topic. After all, you are reading the literature anyway, so why not take the opportunity to note, for example, relevant facts, quotes and supporting evidence for your thesis? Because Lateral is designed to learn from whatever kind of information you’re seeking, this process also fits naturally into the software’s workflow.

References:

  1. Is your brain asleep on the job?: https://www.psychologytoday.com/us/blog/prime-your-gray-cells/201107/is-your-brain-asleep-the-job
  2. Tim Feriss speed reading: https://www.youtube.com/watch?v=ZwEquW_Yij0
  3. Five biggest reading mistakes: https://www.timeshighereducation.com/blog/five-biggest-reading-mistakes-and-how-avoid-them
  4. Skim reading can be bad: https://www.inc.com/jeff-steen/why-summaries-skim-reading-might-be-hurting-your-bottom-line.html
  5. Cryptomnesia: https://en.wikipedia.org/wiki/Cryptomnesia
  6. Systematic literature review with boolean keywords: https://libguides.library.cqu.edu.au/c.php?g=842872&p=6024187

Lit review youtube intro: https://www.youtube.com/watch?v=bNIG4qLuhJA

More in

Research

Ready to go Lateral?
Get new perspectives with Lateral AI.

Get started for free. No credit card required.

By clicking “Agree”, you agree to the storing of cookies on your device to enhance site navigation, analyse site usage, and assist in our marketing efforts. View our Privacy Policy for more information.