HelpMicrosoft Academic Graph in EPPI-ReviewerUpdating a review / 'citation chasing'
Bringing a review up to date and performing ‘citation chasing’

The third scenario to consider is when we want to bring an existing review up to date and performing ‘citation chasing’. In this situation we already have lots of information about the review, including the studies that have been included and excluded. We can use this information to help us find new and related studies efficiently. Before we are able to use these advanced features though, as many records in the review as possible need to be ‘matched’ to their equivalent in Microsoft Academic. Please see the section below on matching records to Microsoft Academic.

Assuming you have the records included in your review matched to their Microsoft Academic equivalents, you can use the features described in this section to bring your review up to date.

First of all, you can use the automatic Boolean search generator[1]  to create a Boolean search for you, based on the studies already included in your review. This function analyses the studies you have previously included and attempts to create a Boolean search that is capable of identifying as many of the records you have included as possible, without including vast numbers of irrelevant records. The algorithm analyses both the titles and abstracts of your included studies as well as the ‘graph’ around your studies: the networks of citations and recommended studies. After running the Boolean search, and importing the new records into your review, you can use the ‘priority screening’ function to rank the new records automatically in terms of which is most relevant to your review.

The second method involves using the ‘graph’, or ‘network’, of publications in Microsoft Academic, all of which are related to one another through their citations, semantics, authors, place of publication, and institutional affiliations. If we start with one or more ‘known’ publications, we can follow their relationships to find other, similar publications. This is sometimes known as ‘snowball’ searching, and while it can be efficient, systematic reviewers tend not to rely completely on these approaches because of potential bias. (e.g. when a relevant document is not cited by other relevant documents) We are currently still evaluating the best approach to take in relation to ‘graph’ searching, but present six possible options, depicted in the figure below.

The above Figure outlines two types of relationships between documents: ‘citation’ and ‘composite graph’. The use of citation relationships has been discussed many times over the years, as it can be an efficient way of finding relevant studies, based on studies that are already ‘known’. It is possible to follow citation relationships in two directions: first, the papers listed in the bibliographies of ‘known’ records; and second, papers which cite the ‘known’ records. In addition, Microsoft has analysed the large number of different ways in which documents can be related to one another, and has created a composite of these relationships known as the ‘related documents’ feature. Any document can have a maximum of 20 other related documents and, again, these relationships can be followed in either direction.

While there are concerns about using citation networks as being a single method of searching, they are a valuable and valid part of a systematic search strategy. As long as your records are ‘matched’ against their Microsoft Academic equivalents, it’s easy to conduct bi-directional citation chasing within EPPI-Reviewer (a task that takes much longer if done manually!).

 

[1] This algorithm is not yet in the user interface, but will be ready soon.

Copyright 2021 by EPPI-Centre :: Privacy Statement :: Terms Of Use :: Site Map :: Login
Home::Help::EPPI-Mapper::RIS Export::About::Account Manager