I hate writing literature reviews, here is how I have automated the process using Zotero, pandoc and Obsidian
Literature Reviews
I hate writing literature reviews. I recently read the article entitled 3+1 Innovative tools that can help you create a Literature review — Research Matrix of papers — Scholarcy, Elicit.org, Lateral.io and Notion/Obsidian/Roam Research… | by Aaron Tay | Mar, 2022 | Medium This article contains a number of tools that can greatly facilitate writing literature reviews and help organize your research. The purpose of the present article is to illustrate in more depth how these tools can be used to organize, and ultimately complete, your literature review in the most efficient way possible.
Software Tools
Creating a Clearly Defined Research Topic
1. Elicit.org, automatically creates a research Matrix of papers using an autoregressive language model that implements deep learning to produce human-like text. The MIT Technology Review called this technology ‘shockingly good’ and ‘completely mindless’ which, if you are like me, is exactly what is needed when writing a literature review OpenAI’s new language generator GPT-3 is shockingly good — and completely mindless | MIT Technology Review
I have a research question in mind (but if you do not, Elicit can help you form one too — Click on Tasks and the “Brainstorm Research Question.” Positive youth development is one of my main research areas. So, in the text box below, I typed: “What is neighborhood approach to child maltreatment?” It brought up a short list of some articles. (Here, I focus on the first few articles that are listed, but you can keep scrolling to download as many as you wish, and can create a custom download of the most interesting articles as well).
One goal is to select articles and have these tools create a bibliography that can be imported into your reference manager (I am using Zotero). I also want to make sure that the pdf attachments are included in my reference manager so that I can have them for further use. Here is how I automated this process. First, click on ‘Add a column’ and select DOI. This step is important because it allows us to download all of the pdf files from Zotero very easily. Without the DOI (and a link to a university library account) there is no guarantee this will work properly).
Elicit does not download the DOI into the exported bibtex file (if it did we would be done because the bibtex file could be directly imported into Zotero). As well, Zotero does not import csv files to my knowledge. So, my workaround was to download my search results as a csv (which does save both the DOI and DOI URL) and then create this bibtex file using the following commands (which I created as a batch file in order to run in Windows using the command prompt).
Download the .csv file (hopefully the next version will allow us to download the DOI with the export)
Delete all columns except for the DOI URL column (delete the first row of column names) and save as a .txt file
The use notepad and type the following (obviously replacing the path). Make sure to save the file as a .bat file.
@echo off
for /f “tokens=*” %%a in (C:\Users\PATH TO TEXT FILE.txt) do (
echo line=%%a
curl -LH “Accept: application/x-bibtex; charset=utf-8” https://doi.org/%%a >> C:\Users\giaba\Downloads\child-malt-result.bib;
)
Assuming that you have Zotero already installed, double click on the bibtex file and import it as a new collection. You should see something like this:
Highlight all entries and right click, select Find Available PDFs
And like that, we have a bibliography associated with our research question with links to full text (assuming they are found, here one was not found).
But, how can we use this information to begin to create a literature review for a research article?
Summarizing papers
2. Scholarcy — automatically summarize papers to extract predefined columns of information
First, browse to the location in which Zotero saves the attachments. We are going to import these files into Scholarcy. You can sign up for a free 14-day trial. Once you log in you should see the following:
Click ‘Create Library’ and name the folder something (I named mine Child Maltreatment). Locate the pdf files that were downloaded from Zotero above. My folder is called ‘C:\Users\giaba\Zotero_Attachments\child-malt-result’ Drag the files into Scholarcy of browse to the folder and select all the files
From the Child Maltreatment folder that we created, click on the first entry and select one of the entries. You will see a number of tabs. My favorite is the ‘Comparative Analysis’ section.
Before we leave this tool, go back to the main folder and select ‘Move/Export/Delete’ –> Export
Make sure all of the articles are selected Change the dropdown to ‘Markdown’ and click ‘Export Selected’
A zipfile will be downloaded called export.md. We will later import these files into Obsidian.
But how can we compare and contrast these documents?
3. Lateral.io, add pdfs of papers and quickly create Research Matrix of papers
Create a free account and then a new project. Mine is titled ‘Child Maltreatment’. Upload the same pdf documents as above.
Once the files are uploaded, you will see something that looks like this
From here, you can create a custom matrix that will automatically be populated based on the fields you create. You can select the snippets that you want added to the matrix. Below I added columns for ‘concepts’ and ‘research questions’.
We want to download the results in markdown format so that we can use the results in Obsidian. Unfortunately, as of now, Lateral.io only exports the fields into word (although it will download the bibliography in bibtex format). So, we will download the results in word and use pandoc to convert the docx to markdown. First, click ‘Export Table’ and then ‘Table as Word.’ Then, convert the word document using the following command
pandoc -s PATH\FILE NAME.docx -t markdown >> child-maltreatment.md
Writing the Literature Review
But how can we begin actually writing the literature review?
4. Obsidian
Navigate to the file folder in your Obsidian “Vault” where and unzip the zipfile that we downloaded from Scholarcy. The unzipped files are in markdown format and can be used in Obsidian. Here is an example output.
Implications for Social Policy
• Public funding for the CARES administrative infrastructure, which was central to organizational development and local capacity building in the original model and funded by the demonstration grant, disappeared with the federal grant.
• Most subsequent funding streams excluded administrative costs.
• Diffusion of a successful innovation requires a commitment to longer-term flexible funding.
• Local-serving organizations cannot institutionalize successful programs, and low-opportunity neighborhoods continue to be repositories of repeated demonstration grants that alone are unlikely to achieve long-term outcomes
Implications for Social Work Practice
• The area was a voluntary-sector-rich locale, confirming [^Wolch_1990_a]) findings that benefits accrue when resources are targeted to an area with local institutional capacity for flexibility and innovation
• This finding has implications for the trend by government to decentralize and privatize formerly public social services.
• Some states are soliciting for-profit and nonprofit contractors to develop “first-to-work” training programs for welfare recipients in targeted areas to implement the recently passed Temporary Assistance to Needy Families Block Grant
• Practitioners involved in such initiatives need to assess and use existing local institutions in the targeted geographic area and cooperate with colleagues employed there to build on family and community development initiatives already under way
Implications for Social Work Research
• The evaluation of complex community-based interventions involving interorganizational collaboration requires sustained field observation and multiple methods of applied, community-oriented research.
• O’Connor (1995) and [^Weiss_1995_a]) argued that the evaluation of complex neighborhood initiatives requires methods that are both quantitative and qualitative, contextual and theory based.
• Most service NPOs rely on government support for more than half of their revenues ([^Smith_1993_a]).
• This dependency is a potential problem for neighborhood-based initiatives in which the locus of management decisions should rest at the community level.
• Further research is needed to help untangle these perplexing issues.
Builds on previous research
• The project developed a complex neighborhood service network of complementary programs that are described elsewhere ([^Mulroy_1997_a]). The example considered here was institutionalized into its target neighborhood during the five-year grant period, and an adapted model was replicated in two other poverty neighborhoods
Future work
• More research is needed that examines the contracting relationship between government and NPOs that has heightened concern about nonprofit autonomy ([^Ostrander_1987_a]; [^Smith_1993_a]; [^Wolch_1990_a]). Most service NPOs rely on government support for more than half of their revenues ([^Smith_1993_a]). This dependency is a potential problem for neighborhood-based initiatives in which the locus of management decisions should rest at the community level. [^Wolch_1990_a]) suggested that community-based agencies may not be able to remain autonomous when targets of change are likely public agencies that are existing or potential funders. Fiscal resource dependencies that push NPOs toward entrepreneurial behavior and collaboration may create conditions under which competition undermines cooperative behavior ([^Alter_1993_a]; [^Mulroy_1997_a]). Further research is needed to help untangle these perplexing issues.