doc.txt

The "Serverless Literature Dataset" (3rd generation)

History:
--------
In February 2018, we have published the first iteration of the SLD
online at https://zenodo.org/record/1175424. It contained 34 works
retrieved via DBLP and organised in a spreadsheet. In September 2018, we
extended it with more recent publication, yielding 44 works. Due to
additional metadata, we quickly hit the limit of spreadsheet
representation and resorted to this more sophisticated set of JSON files
which allow for easier processing and textual change tracking. In late
September 2018, we have published the second iteration of the SLD with
60 works at https://zenodo.org/record/1436432, and published a paper
about the dataset at http://ceur-ws.org/Vol-2330/short2.pdf. In April
2019, we have published the third dataset... and in October 2019, the
fourth one.

Read more:
https://blog.zhaw.ch/splab/2019/04/22/the-5th-year-of-serverless-computing-research-coverage/
http://serverless.research-output.org/

Citation:
---------
SLD contains a script 'make-bib.py' which produces a file
'serverless.bib' with stable identifiers. Add this file with
\addbibresource{serverless.bib} to your biblatex setup.

You can also cite SLD itself as follows:

 @misc{sldataset,
 title = {{Serverless Literature Dataset}},
 author = {Josef Spillner and Mohammed Al-Ameen and Daiana Boruta},
 howpublished = {Zenodo dataset (4th revision) at \url{https://doi.org/10.5281/zenodo.1175423}},
 year = {2019},
 month = {October},
 doi = {https://doi.org/10.5281/zenodo.1175423}
 }

SLD is a collaborative effort - we call on all researchers in the field
of serverless computing to extend and augment the data as the research
evolves! Significant contributions will be credited with co-authorship.

Installation:
-------------
This step along with the subsequent ones is only required if you want to
contribute to the dataset and extend it in a consistent way.

- % sudo apt-get install python3-pybtex # alt.: pip3 install pybtex
- % sudo apt-get install python3-matplotlib-venn # alt.: pip3 ...
- % sudo apt-get install poppler-utils graphviz

Usage:
------
The basic curation workflow is as follows:
- find DOIs, arXiv and USENIX identifiers via e.g. DBLP ("faas",
  "serverless (application/computing)", "function-as-a-service",
  "lambda", "cloud function")
- use scraper/parsedblp.py to assist with this process
- otherwise, add a DOI, arXiv or USENIX identifier manually to
  serverless-literature-base.json
- run populate.py to produce serverless-literature-bibliography.json
  from this file (see comments in populate.py for usage)
- if no DOI or arXiv links are available, add relevant fields including
  "link" directly to serverless-literature-bibliography.json and leave
  an empty entry with "link" in the base file
- commit the results

Ensure that each key in serverless-literature-base.json is unique! All
other checks, in particular consecutiveness, will be checked for by the
tool. Relevant fields are "title", "author", "year", and "journal"/
"booktitle".

The advanced curation workflow is as follows:
- run analysischeck.py and analysistitles.py
- follow the advice, extend serverless-literature-analysis.json
  and serverless-literature-technologies.json appropriately
- look at curation.txt for hints on filling the values
- compare with https://www.tu-berlin.de/?id=199198 to see if any FaaS
  benchmarks exist that should be crosslinked into benchmarking.json

Once all data has been checked, statistics output can be produced as by
the following instructions:
- run stats.py
  -> produces a file stats.txt with overall statistics
- run tagcloud.py (if folder 'pdfs/' with files '<number>.pdf' exists)
  -> produces a file _tagcloud.txt to build tag clouds with
- run venn.py
  -> shows several Venn diagrams highlighting search terms in titles
- run venue.py
  -> shows text output on where papers have been published
- run graph.py
  -> plots a PDF file with researcher-technology mappings
- run make-bib.py
  -> produces a citable file serverless.bib with key bibliographic data