LitRevPro (Literature Review Project) is a collection of small lightweight apps to help academics conduct a systematic literature review using AI.

Current working tools

  1. Dataset parser – prepare Web of Science records for manual classifier
  2. Manual classifier – create training data set for AI algorithm
  3. Automative binary classifier – apply classification algorithm

Where is comes from

I find myself conducting literature reviews quite often. When doing so, one of the critical (and most painful) steps is identifying and selecting the relevant academic articles to include in the study.

This has become an even bigger problem due to the increased accessibility to academic databases such as World of Science and Scopus. These databases make it easy to search and find a vast amount of literature on any given subject in a really short time.

The increased amount of data, however, can make it really difficult to sift through all the academic papers and identify the important ones.

In other words, how can you sift through all those articles to find the ones you really really want?

Can AI help?

The quick answer to this is “Sure, why not?”

AI and machine learning techniques like classification and semantics analysis can help classify a large volume of articles in a short period of time. But what we really need to know is – when and how?

With regards to “when” we need to know what type of literature review we are conducting. For instance, Cronin et al. (2008) identify 4 types of literature reviews, each differing in terms of purpose, scope, and methodological approach.

The one I am most interested in is the systematic literature review, where the scope and methodology play an even bigger role. Conducting a systematic literature review requires several stages of selection, classification, and exclusion/inclusion (Corin et al., 2008; Shahrivar et al., 2018).

The next question is focused more on AI / machine learning itself. What is it?

Like all technological fads, it seems that everybody likes to talk about machine learning and AI like they know what they are talking about. It is still important to understand what it is and what it is not, so we can leverage its strengths and understand its limitations.

Without getting into a technical or philosophical discussion as to the definition of AI and machine learning, from a practical perspective I will describe AI and machine learning as a group of algorithms and approaches that facilitate the evaluation and analysis of large amounts of data (remember Big Data?).

The actual algorithms involved may have a wide range of functions and specialisations. For the purpose of the literature review their function will focus on classification. The classification can be binary (yes or no) or non-binary (option a, option b, option c). It can also deal with similarities between document based on numerical data…more on that later.

Where LitRevPro fits in

LitRevPro aims to incorporate machine learning algorithms into a cohesive group of apps to conduct a systematic literature review.

At this very early stage there are three software tools that have already been made:

1 – Dataset parser

This stool, written in Python 3, parses a World of Science exported search results and prepares it for the binary classifier.

[more information on the dataset parser can be found here]

2 – Manual Classifier

The first tool provides an easy to use interface for creating a training set to be used for the binary classification.

3 – Automatic binary classifier

The binary classifier then takes over the training set and helps identify related articles based on their abstract.


Cronin, P., Ryan, F., & Coughlan, M. (2008). Undertaking a literature review: a step-by-step approach. British journal of nursing, 17(1), 38-43.

Shahrivar, S., Elahi, S., Hassanzadeh, A., & Montazer, G. (2018). A business model for commercial open source software: A systematic literature review. Information and Software Technology.