1. SCIENTIFIC
PAPER WRITING
Computer Science Department
Universidad Autónoma de Madrid (Spain)
http://miso.es
@miso_uam
Juan de Lara
Madrid – June 8th 2020
This project has received funding from the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie grant agreement n° 813884.
2. AGENDA
Part I: Scientific publication
Why publishing?
Types of venues, review processes, types of papers
Ethics, rankings
Part II: Scientific writing: the literature part
How to organize an article?
Some tips
Part III: Scientific writing: the technical part
LaTeX, BiBTeX, Overleaf
Part IV: Scientific writing & reviewing: the practical part
Reviewing, Easychair
Our writing workshop
2
3. SCIENTIFIC
PUBLICATION: WHY?
Communicate science
• Makes scientific progress possible
• Allows researchers to build on each other results
Should be rigorous
• Peer-reviewed
• Very different from other types of literature
• Very different value and trust than a post on a blog
Worthy
• Each paper should make a contribution
• Ensure novelty and progress
3
4. SCIENTIFIC PUBLICATION
DURING A PHD: WHY?
Shows evidence of PhD novelty
Peer-reviewed publications as a prerequisite for PhD defense
• Way to ensure a minimum of PhD quality
• Publications as proxies for relevant work
• In our department (UAM): two journal publications
Get involved with the community
• Get feedback from other researchers
• Present the work at conference, discuss, exchange ideas
• Other researchers can build on your results
Essential skill for a successful academic career
4
5. TYPES OF VENUES
Journals
• Specialized technical content
• Normally unlimited length for papers
• Publication typically takes > 1 year
• Top in our area: IEEE Transactions on Software Engineering (for
SE), SoSyM
Magazines
• Shorter papers, more informal style
• Typically for a broader audience, sometimes professionals (not only
academics)
• Top in our area: Communications of the ACM, IEEE Software
Books
• Monographs vs. collections (eg., result of seminars, set of papers
on a topic)
• Typically specialized, sometimes for teaching (e.g., post-graduate)
• Top book publisher in our area: Springer
5
6. Conferences
• On a particular topic
• Acceptance rate, limited length
• Quicker publication compared to journals (answer in a couple of
months)
• Conference registration is not free, access to proceedings is
typically not free (some open access initiatives, like ETAPS)
• Yearly basis, strict deadlines
• Top in our area: MODELS, ICSE (for SE)
Workshops
• On a very specialized (emerging, hot) topic
• Affiliated to conferences
• More permisive, the goal is discussion
6
TYPES OF VENUES
7. PUBLICATION ACCESS
Open access
• Diamond/Platinum: free of fees and access charges (e.g.,
JOT, http://www.jot.fm)
• Gold open access: free for readers. Some publishers charge
article processing charges (APCs) or fees (APFs) (e.g., IEEE
Access).
• Hybrid: closed journals, where authors can pay a fee to make
their papers open. Most journals nowadays.
Closed
• Free for authors, paying for readers
• Institutional subscription
Green access
• Some publishers allow authors self-archive their publication
pre-print (e.g., in arXiv, institutional repository, web page)
7
8. THE PUBLICATION
PROCESS
1. Work has been done that is worth publishing:
• Improves something relevant (e.g, a more efficient tool)
2. An appropriate venue is selected
• Workshop for incipient works
• Conferences for more mature works, include some evaluation
• Journal for finished works, including extensive evaluation
3. The paper is (carefully) written
• What is in the paper is the only way your work is to be evaluated
• Don’t expect to write a paper in one day, or even one week
• You should position you work w.r.t. related research, motivate the
need for your work and demonstrate that your work improves the
state of the art (evaluations)
4. The paper is submitted
(not done yet!...)
8
9. 9
4. The paper is reviewed
• Anonymous reviewers (other researchers)
• Suggestion for changes and acceptance notification
5. Notification
• Accept/Reject for conferences [sometimes rebuttal]
• Reject/Major/Minor changes/Accept for journals
[Major/Minor changes]
1. Paper revision
• Improve the paper with the reviewers’ comments
• Prepare a letter explaining how you changed the paper and answering the
reviewers’ comments
2. Submit revision and Go-To 4
[Accepted]
1. Preparation of Camera Ready
• No need to prepare a letter, just submit the sources (e.g., LaTeX)
2. Revision of final paper proofs
• As prepared by the publisher
THE PUBLICATION PROCESS (CONT.)
10. THE PUBLICATION PROCESS:
PEER REVIEWING
Single-blind
• Program Committee for Conferences (subreviewers)
• Editorial Board for Journals (reviewers)
• But reviewing is anonymous: you do not know who has reviewed your
paper
• Authors are not anonymous
Double-blind
• Authors are also anonymous (i.e., names do not appear in the paper
itself)
• Avoids bias
• Becoming the norm nowadays for conferences (e.g., ICSE, MODELS)
Open
• Authors and reviewers are known by all participants.
10
11. TYPICAL EVALUATION
CRITERIA (ICSE’21)
Soundness
• The extent to which the paper’s contributions are supported by rigorous
application of appropriate research methods
Significance
• The extent to which the paper’s contributions are important with respect to
open software engineering challenges
Novelty
• The extent to which the contribution is sufficiently original and is clearly
explained with respect to the state-of-the-art
Verifiability
• The extent to which the paper includes sufficient information to support
independent verification or replication of the paper’s claimed contributions
Presentation:
• The extent to which the paper’s quality of writing meets the high standards of
ICSE, including clear descriptions and explanations, adequate use of the
English language, absence of major ambiguity, clearly readable figures and
tables, and adherence to format
11
12. PUBLICATION PROCESS:
ETHICS (AS AN AUTHOR)
It is not allowed to submit to the same work to several venues
• Each paper should contain substantial improvement to previous
publications. Proper citations.
• It is normal to extend conference papers to journals (>30% increment)
It is not allowed to copy phrases from other papers (plagiarism)
• Not even from your previous papers (self-plagiarism)
• Proper citation, also for images (request premission)
It is not allowed to fabricate experimental results
• Scientific fraud, criminal consequences
• Make data openly available
Smell: many authors in publications
• “pyramid scam”
• Each author should have contributed to the paper
• Order of authors
12
13. PUBLICATIONS:
WHERE?
Publisher data bases
• IEEE Xplorer
• ACM Digital Library
• Elsevier’s scopus
• Springer Link
• (most require subscription)
Community data bases
• DBLP
Others
• Researchgate (SN-like)
• Sci-hub (illegal?)
13
14. RANKINGS (VENUES)
Not every journal or conference has the same quality
Conferences
• CORE (http://portal.core.edu.au/conf-ranks/)
• Acceptance ratios, reputation
• A* (ICSE, OOPSLA) – acceptance rates around 10-15%
• A (MODELS, ASE, CAISE) – acceptance rates around 20-25%
• B (FASE, SLE, SEAA)
• C, no rating (ECMFA)
Journals
• Journal Citation Report (JCR)
• Impact based on how many citations each published paper receives
• The higher the impact factor, the higher relevance each paper has
• In our area: IEEE TSE, ACM TOSEM, Empirical Software
Engineering Journal, SoSyM, IEEE SW, IST, JSS
• Classification in quartiles (Q1, Q2, Q3, Q4)
14
15. RANKINGS
Beware of predatory journals and conferences
They are just a lucrative business
• No real quality assessment of papers
• You’ll receive a lot of SPAM about these journals and conferences
• Never submit!
• Will harm your CV and your reputation
• https://predatoryjournals.com/journals/
15
16. RANKINGS: AUTHORS
Number of publications
• Quantity quality
Number of citations
• How relevant is the work
• Depends on the area (some areas, like ML/AI publish much
more, and much shorter papers)
h-index
• h-index = X if you have X papers cited at least X times
16
17. RANKINGS: WHERE?
Google Scholar
• Publications, citations
• h-index
DBLP
• Publications
Scopus
• Publications, citations
• h-index
Web of Science (non-free)
• JCR
Microsoft academic
• Venues, authors
Others
• Semantic scholar, researchgate
17
18. TYPES OF PAPERS
Systematic Mapping Study
• Analysis of the literature of a given (hot, narrow) topic
• Understand what is the state of the art
• Map the frequencies of publication over time to see trends
• Useful for newcomers into the area
• Requires high effort
• Main venues: ACM Computing surveys, IEEE TSE, SoSyM, IST, etc
18
19. TYPES OF PAPERS
Research paper
• Describes research work done
• Should compare with related research (not to the detail of an SMS!)
• Long, full, regular: completed research, including evaluation
• Short: new ideas, work-in progress, workshops
19
20. TYPES OF PAPERS
Experience report
• Challenges faced, approaches
taken, observations, and
insights when applying a
technique to a real problem
• Many times linked to application
of research to industry
• Industrial track in many
conferences
• P&I track at MODELS, ECMFA,
ICSE, etc
20
21. TYPES OF PAPERS
Empirical evaluation
• Measure efficiency/effectiveness of a technique
• User’s study (e.g., whether tool X is better than tool Y when
performing an activity)
• Empirical software engineering
• Main venues: EMSE Journal, ESEM conference, EASE conference
21
R. Ren, J. W. Castro, A. Santos, S. Pérez-Soler, S. Teresita Acuña, J. de Lara: Collaborative
Modelling: Chatbots or On-Line Tools? An Experimental Study. EASE 2020: 260-269
22. TYPES OF PAPERS
Tool demos
• Short paper describing a tool
• Most conferences have a tool demo track
• Typically a separate track in the conference
• The idea is to make a demo during the conference
• Sometimes a video is required together with the paper
22
Cuadrado, Guerra, de Lara. AnATLyzer: an advanced IDE for ATL model transformations.
ICSE (Companion Volume) 2018: 85-88
23. TYPES OF PAPERS
Poster
• Very short paper describing an incipient
idea
• Not presented via a talk, but
summarized with a poster
• Read by other participants in coffee
breaks
• Author answers questions of interested
people
23
24. TIPS
Be prepared for rejections
• Failure is part of the process, but you need to learn from it
• Take advantage of the reviewers’ comments to make a better paper
• If you cannot cope with rejections, an academic career is not for
you: consider leaving the PhD
• Everyone has rejected papers
Take ownership of your work
• Do not expect that your supervisor will write the paper for you
Good research needs good writing
• Good writing skills are very valuable in a researcher
Inspiration exists, but it has to find you working [P. Picasso]
• Everyone can have good ideas, but need to be realized
• A good paper requires a tremendous amount of work
24
25. AGENDA
Part I: Scientific publication
Why publishing?
Types of venues, review processes, types of papers
Ethics, rankings
Part II: Scientific writing: the literature part
Dissecting an article
Some tips
Part III: Scientific writing: the technical part
LaTeX, BiBTeX, Overleaf
Part IV: Scientific writing & reviewing: the practical part
Reviewing, Easychair
Our writing workshop
25
26. DISSECTING AN ARTICLE
Title
Abstract
Keywords and classification terms
1. Introduction
• Sometimes another section on motivation, otherwise included
in introduction
2. Sections describing the approach (1+)
• Typically two or more sections (overview, concepts)
• Section on architecture and tool support
• Section on evaluation
3. Related Work
• Sometimes after introduction, to motivate the lack of works
4. Conclusions and future work
References
26
27. DISSECTING
AN ARTICLE
Title
• Should concisely convey what the paper is about
• Avoid long titles
• Catchy to call the attention of reviewers and readers
• Sometimes a subtitle
• Examples: “Datalyzer: Streaming Data Applications Made Easy”,
“Refactoring Multi-Level Models”, “Facet-oriented modelling: open
objects for model-driven engineering”
Authors
• Position is important (1st author: most of the contributions)
• In our department, PhD students are required to have 2
(journal) publications as first authors
• Alphabetical order (depends on research group)
• Each author should have contributed to the work described in the
paper in some way (ideas, implementation, evaluations, writing, etc)
27
28. DISSECTING
AN ARTICLE
Abstract
• Summarizes the main points of the paper
• Short and to the point, should incite the reader to continue
reading
• Sometimes limited in length
• Structured abstracts (often found in empirical software
engineering venues):
• Context
• Objective
• Method
• Results
• Conclusions
28
30. KEYWORDS AND
CLASSIFICATION TERMS
Typically 3+ keywords characterizing your work
• From general to more specific, like: Model-driven engineering,
model transformations, ATL
• Help to assign reviewers, and improve search
ACM Classification terms (e.g. ACM)
• Hierarchy of standard computer science topics
• https://www.acm.org/publications/class-2012
• Example:
30
31. INTRODUCTION (1/2)
General goal: introduce the problem you are solving, motivate the
work, describe what is in the paper, and its main contributions
“funnel shape”: broad at the start (general background of the
problem) and narrow at the end (aim of current paper) [1]
Perhaps the most important section of the paper
• Entices the reader to keep reading
• Engages the reviewer, and persuades her to a good score
Puts the work in context
• May be some short background
• Depends on the audience you are presenting the paper to
Introduces the current situation
• Presents a problem you aim to solve
• Why is it relevant?
31
[1] http://www.scientificwritingtips.com/
32. INTRODUCTION (2/2)
Shortly describes the paper contribution
• Should attack the problem just presented
• Possibly enumerate the paper contributions
For incremental work, explain the delta
• Always explain the advances with respect to your previous work
• Omision of this can be a reason to reject
For short papers, sometimes related work is embedded here
Finishes with a paragraph describing the organization of the rest of
the paper
• “The rest of this paper is organized as follows. Section 2, ….”
32
34. MAIN SECTIONS
Main part of the paper
May be useful to start with an overview section
• Describe usage scenarios
• Provide a general scheme
Architecture and tool support
• Very common in SE papers
Evaluation
• Many conferences, and all journal papers should provide
some sort of evaluation
• Under some metrics: performance, precision and recall, etc
• Formulation and answering of research questions (RQs)
• User evaluation for tools
34
36. RELATED WORK
Compare with existing works in the literature
• Position your work with respect to previous ones
• Quite exhaustive, but not like an SMS
• Explain what is novel in your work with respect to these
approaches
• Failure to identify relevant related work may imply paper
rejection
• Group related work on several aspects into subsections
• Sometimes finish with a paragraph summarizing the novelty
of your work with respect to all analysed related work
36
38. APPENDICES
Extra information and details that would interrupt the reading
flow:
• Proofs of theorems in the paper
• Listings
• Forms used in an empirical evaluation
• Additional experimental data
Example: http://miso.es/pubs/TOSEM-trm.pdf
38
39. TIPS (1/2)
Follow a clear argumentation line
• Approach writing as you would write an algorithm
• Know where you are heading to, take the reader there step by
step
• Do not make large leaps in your reasonings
• Document each claim in your reasonings with references
Write succinctly, go the the point
• Less is more
• Avoid long phrases
• Avoid long paragraphs
• Don’t confuse the reader with long phrases
39
40. TIPS (2/2)
Make clear what is the paper contribution
• Write it down explicitly
• Don’t expect the reviewer or the reader to figure it out by
themselves
• Avoid papers of the kind “I’ve don’t this”
Show that you are familiar with related work, and explain
what is novel in your approach
• Novel in results, not in the technique employed
• For example: solve a solved problem with MDE is not a
contribution. Why is the solution better?
Make proper use of
• Images, including diagrams, graphics, screenshots
• Tables
40
42. AGENDA
Part I: Scientific publication
Why publishing?
Types of venues, review processes, types of papers
Ethics, rankings
Part II: Scientific writing: the literature part
Dissecting an article
Some tips
Part III: Scientific writing: the technical part
LaTeX, BiBTeX, Overleaf
Part IV: Scientific writing & reviewing: the practical part
Reviewing, Easychair
Our writing workshop
42
43. WRITING THE PAPER
No one uses Word
• Not professional results
• Often perceived as a bad sign of quality
• Difficult for collaborative writing
• Difficult for version control
• Bottom line: don’t use Word for writing papers
LaTeX is the norm
• Similar to HTML, or to code
• Compiled into pdf (pdflatex)
• Also possible to compile to .dvi (latex) and then to .ps (dvips),
but try to avoid that
43
44. HELLO WORLD
IN LATEX
Create tex file article.tex, with content:
Compile with: pdflatex article
Open article.pdf
44
documentclass[a4paper,10pt]{article}
begin{document}
Hello world! % Writes Hellow world!
end{document}
Thanks to: http://www.docs.is.ed.ac.uk/skills/documents/3722/3722-2014.pdf
45. HELLO WORLD
IN LATEX
Anything starting by is a LaTeX command
• Commands may have parameters
• You can declare your own
documentclass must appear at the start of every document
• Article is the type of document (other possibilities report, proc,
book, slides)
The text of the document goes between begin{document}
and end{document}
• Preamble: everything before begin{document}
You may need to compile twice to resolve references
45
47. ADDING SECTIONS
47
documentclass[a4paper,10pt]{article}
begin{document}
title{My First Paper}
author{Your name here}
date{today}
maketitle
section{Introduction}label{sec:intro}
We will finish in Section~ref{sec:conclusions}.
section{Conclusions and Future work}
label{sec:conclusions}
end{document}
Subsections with subsection, subsubsection
label to produce a referenciable anchor (via ref{})
49. ADDING THE
ABSTRACT
49
documentclass[a4paper,10pt]{article}
begin{document}
title{My First Paper}
author{Your name here}
date{today}
maketitle
begin{abstract}
A short {em abstract} comes {bf here}.
end{abstract}
section{Introduction}label{sec:intro}
We will finish in Section~ref{sec:conclusions}.
section{Conclusions and Future work}
label{sec:conclusions}
end{document}
51. LISTS
51
documentclass[a4paper,10pt]{article}
begin{document}
% same as before
section{Introduction}label{sec:intro}
An enumeration of items
begin{enumerate}
item One item in the list
item Another item in the list
end{enumerate}
An itemized list
begin{itemize}
item One item in the list
item Another item in the list
end{itemize}
end{document}
53. PAPER TEMPLATES
Most conferences and journals require using a certain
template, to ensure uniformity
• ACM, IEEE, LNCS, Elsevier, Springer are the most common
Download the template files and add them to your paper
folder
Just a documentclass parameter
• documentclass{llncs}
• documentclass[sigconf]{acmart}
• documentclass[10pt,journal,compsoc]{IEEEtran}
Templates also bring commands
53
57. HANDLING
BIBLIOGRAPHY
BibTex: http://www.bibtex.org/
“Data-base” of paper meta-data
• File article.bib
• Specific format
Specific commands in the text to produce citations
• cite{paperName}
You need to compile with: bibtex article
• Several cycles of pdflatex / bibtex may be required
• Compiling with bibtex produces a .bbl file (never edit that
directly)
Other options
• Biber
57
58. BIBTEX FORMAT
(CONFERENCE PAPERS)
@inproceedings{AroraSBZ16,
author = {Chetan Arora and Mehrdad Sabetzadeh and Lionel C.
Briand and Frank Zimmer},
title = {Extracting domain models from natural-language
requirements: {A}pproach and industrial evaluation},
booktitle = {Proc. MODELS},
pages = {250--260},
year = {2016},
publisher = {{ACM}},
}
58
59. BIBTEX FORMAT
(JOURNAL PAPERS)
@article{DalpiazFFP18,
author = {Fabiano Dalpiaz and
Alessio Ferrari and
Xavier Franch and
Cristina Palomares},
title = {Natural Language Processing for Requirements
Engineering: The Best Is Yet to Come},
journal = {{IEEE} Software},
volume = {35},
number = {5},
pages = {115--119},
year = {2018},
}
59
60. REFERENCE FORMATS
Normally different paper templates will produce different
reference formats
• References numbered by order of appearance
• References numbered by alphabetical order
• Authors’ initials instead of numbers: [LGS20]
• Authors’ names instead of numbers: (Dalpiaz et al., 2020)
No need to change anything, except selecting the appropriate
format
60
61. TIPS
Break the paper into files
• One per section
Put images in a separate folder
Create folders for: sources of images, auxiliary files
Use vector graphics for images
• In pdf format when possible
Create macros for frequently used elements
• Figures, mathematical formulae, etc
Number different parts of screenshots
• Avoid “the upper-right corner of the figure…”
• Better “label (1) in the figure…”
Never submit without passing a spellchecker
Use lstlisting package for listings
61
62. LATEX IN PRACTICE
Many editors
• A basic one + command line
• TeXworks (many others)
Collaborative editing
• Put your paper on a control version system like github
• Best option
• Conflicts in text are easy to handle
• Allows off-line work
• NEVER put .bbl, .aux, .log, article.pdf under version control
• Use Overleaf (google docs-like collaboration)
• You will not see what other people are doing
• Requires being on-line
• Good for non-techies
62
64. AGENDA
Part I: Scientific publication
Why publishing?
Types of venues, review processes, types of papers
Ethics, rankings
Part II: Scientific writing: the literature part
Dissecting an article
Some tips
Part III: Scientific writing: the technical part
LaTeX, BiBTeX, Overleaf
Part IV: Scientific writing & reviewing: the practical part
Reviewing, Easychair
Our writing workshop
64
65. REVIEWING SYSTEMS
Web applications were paper submissions are handled
• Easychair, HotCRP for conferences
• Manuscript central for journals
Support the paper life-cycle
• Submit the paper
• Assign reviewers
• Upload reviews
• Notify authors
• Submit revisions
• etc
65
69. REVIEWING
Carefully read the paper
A review is made of
• A score [-3..3] (strong reject to strong accept)
• A confidence
• The review text
The review text normally includes
• A summary (shows that you’ve understood the paper)
• Positive and negative points on the paper
• Major concerns
• Minor concerns, typos
Always use a positive tone, and include suggestions for
improvement
69
70. PUBLICATION PROCESS:
ETHICS (AS A REVIEWER)
Constructive reviews
• Don’t take advantage of being
anonymous
• It’s not your paper
• No need to be rude
Give proper attention to the paper
• Carefully read the paper
• Prepare a detailed review, as you’d
like to receive for your own paper
Don’t use reviewing to your benefit
• Suggest citing your papers (only if it
is appropriate)
70
71. OUR WRITING
WORKSHOP
We will emulate the whole process of a workshop
You’ll need to:
• Write a paper
• Review each other papers
• Revise the paper
The goal is to
• Learn to write a paper
• Learn to write a review
• Write the draft of your LowCode’20 workshop submission
Of course the outcome of this writing workshop does not ensure
acceptance at LowComote’20
• For the LowComote’20 you’ll probably receive help from your
supervisors
71