Writing a Scientific Article, part II
Vladimír Ulman, Igor Peterlík, Jan Obdržálek
FI MU
DUVOD, November, 2015
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 0 / 27
Technical Aspects of Scientific Articles
English
Algorithms
Figures and charts
Tables
Bibliography
Acknowledgement
Short CVs
Supplementary materials
English and general typesetting (symbols, numbers, units)
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 1 / 27
Scientific English
Science writing is much easier than it looks.
find a good book, for example: Science research writing for
Non-native speakers of English by Hilary Glasman-Deal (free)
14 Science Research Writing
FA
this idea; they discuss it, describe it, define it in more detail, argue about it,
give examples of it, rephrase it, etc. When the ‘topic’ or idea moves too far
1. READ THE TITLE
and try to predict the type of information you expect to see
2. LOOK AT THE NAME OF THE AUTHOR
What you know about the writer will help you predict and
evaluate the content.
3. CHECK THE DATE
and use it to help you assess the content.
4. READ THE ABSTRACT
to find out what the researchers did and/or what they found
5. LOOK QUICKLY AT THE FIRST PARAGRAPH
without trying to understand all the words.
6. LOOK QUICKLY AT THE FIRST SENTENCE OF EACH
PARAGRAPH
without trying to understand all the words
7. LOOK QUICKLY AT EACH FIGURE/TABLE AND READ ITS
TITLE
to try and find out what type of visual data is included
8. READ THE LAST PARAGRAPH
especially if it has a subtitle like ‘Summary’ or ‘Conclusion’
2 Science Research Writing
FA
RESULTS
(what you found/saw)
METHODOLOGY
(what you did/used)
ABSTRACT
INTRODUCTION
DISCUSSION/
CONCLUSION
central
report
section
Fig. 1. The shape of a research article or thesis.
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 2 / 27
Scientific English: Tenses
Present simple
The most important and frequently used tense in scientific writing
Stating accepted facts and truths, permanent situation
(introduction, s.o.t.a.)
Describing standard procedures (methods)
Reported findings are considered as general, another scientist will
necessarily get the same results (results)
Present continuous
Temporary situation (a change is expected) (intro)
Past simple
What you did yourself in contrast to standard procedures (methods)
Indicates that findings are linked only to your research (results)
Present perfect
The event reported is relevant to the actual situation (intro)
“Simple future” (“will”)
To be avoided everywhere except in future work
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 3 / 27
Scientific English: Examples
Correctness of the algorithm was demonstrated two years ago.
However, little attention has been paid to other aspects such as. . .
vs.
However, little attention was paid to other aspects such as. . .
A global minimum was found using a genetic algorithm...
vs.
A global minimum is found using a genetic algorithm...
We found that the convergence of the iterative method strongly
depended on. . .
vs. We found that the convergence of the iterative methods
strongly depends on. . .
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 4 / 27
Scientific English: Sentence Connection
The space between a full stop and the next capital letter is a
dangerous space for you and your reader.
Therefore, use “connectors”
They considerably improve fluidity and facilitate understanding
Cause: since, as, because, due to, on account of
Result: therefore, hence, consequently, so, which is why
Contrast: however, whereas, but, while, by contrast, yet
Unexpectedness: although, even though, despite,
notwithstanding, regardless of, yet, nevertheless, nonetheless
Addition: moreover, further, furthermore, also, secondly, in
addition
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 5 / 27
Scientific English: Talking about Results.
Results do not speak for themselves!
Sequence: order of things (extremely important in algorithms, but
also on a “global level”). Useful words: first, second, then,
afterwards, next, finally, prior to, subsequently, immediately. . .
Frequency: how often an event or result occurred. Useful words:
each time, always, without exception, constantly, repeatedly,
frequently, sometimes, at times, occasionally, from time to time,
rarely, seldom, hardly ever, barely ever, at no time, never. . .
Quantity: great deal, important, considerable, significant,
substantial, appreciable, plenty, reasonable, numerous,
somewhat, particular, moderate, few, little, small, negligible,
exceptional, infinitesimal, marginal
Causality: relationships and connections between observations.
Useful: cause, produce, originate, be due to, initiate, be related to,
be linked to, contribute to, correlates with, be a factor in, results
from, yield, stem from
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 6 / 27
Scientific English: Discussion/Conclusion.
Usually using a lot of “modal verbs“:
Ability/Capability: can, cannot, could, could have, could not. . .
Possibility/Options: may, might, could, may have, might have. . .
Probability/Belief/Expectation: should, ought to, should have,
ought to have. . .
Virtual certainty: must, have to, cannot, must have. . .
Advice: should, ought to. . .
Necessity: must, need to, have to. . .
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 7 / 27
Choosing the Right Template
Conferences usually provide templates, styles for contributions
LATEX templates (LNCS, IEEE, Elsevier)
MS Word templates (mainly in non-CS domains)
The style usually determines:
Page size, columns, headers, footers
Title, captions placement and numbering
Font type (text, captions)
Citation style
The latex template often provides several “modes”
Mainly used in journal submissions (conference is usually in
camera-ready template from the beginning due to the strict length)
Draft, review, final, one-column, two-column
Some properties of the submission can be changeable (mainly in
journals)
Number of pages; buying extra pages often possible
Black/white vs. color
Discuss the financial resources with your supervisor.
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 8 / 27
Tables, Figures and Algorithms
Should increase the comprehensibility of the text
Every object must be referenced in the text
In LATEX use label, ref
The label is generated by the caption command, so label should
be placed after caption to have correct references
Correct referencing:
“. . . in Table 1. . . ”, LATEX : in Table~\ref{tab:label}
“. . . in Fig. 2. . . ”, LATEX : in Fig.~\ref{fig:label}
Placement on the page can be given by template or
recommended by the conference/journal web.
In LATEX Here, Top, Bottom (forcing with with ’!’)
Placement in the top can sometimes save some place (no top
margin needed to separate from the text
In journal submissions, tables, figures and algorithms may be
added to the text separately (mainly in the first submission).
In some journals, paper subsections (state of the art) can appear
in highlighted boxes.
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 9 / 27
Typesetting Algorithms
Surprisingly not very often even in computer science papers
Maybe because of the idea say enough but not too much...?
Typesetting an algorithm should be considered carefully (and
probably avoided in Word-like submissions)
A text description is sometimes sufficient
(but increases dense prose)
Usually quite useful in thesis or longer reports
Several LATEX packages exist for algorithmic environment
Algorithmic, algorithmicx, algorithm2e
http://en.wikibooks.org/wiki/LaTeX/Algorithms_
and_Pseudocode
General rules
Show only essential and rather short functions/procedures
Rely on indentation (usually provided by the package)
Reduce (or avoid) { } delimiting blocks
Reduce the occurrence of special symbols and long variable names
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 10 / 27
Figures and Images
A (good) picture is worth a thousand words.
A figure: drawing, shape, visual representation
Typically a diagram (flowchart, function graph, data chart). . .
Can be very practical when giving the big picture
. . . or a graph (pie, bar, line charts)
Often necessary to present/illustrate the results
An image: a picture
Taken by a camera, extracted from a video, screen shot
Can be very practical to illustrate the results or display “realia”
Consider the requirements for the paper
Color, length, page size, style. . .
A figure can significantly shorten the text (simplified explanations)
Color limitation (b/w) can be an issue
Consider different plotting patterns (dotted lines, thickness,
gray-scale etc.)
Do not forget legend to interpret colors, lines and other objects in
the figure
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 11 / 27
Figures and Images
Employ a vector representation as much as possible (figures)
If pixel representation is unavoidable, keep a sufficient resolution
(600 DPI = 600 pixels per 2.54 cm), convert to pixels if needed
Avoid JPG if the paper is to be printed (not always possible...)
Never use a figure (schemes, graphs, plots) that was saved in
JPG in some phase of processing (blurred edges)
Always consider readability
Contrast, color saturation, size of details (paper cannot be zoomed)
Font size if any text or symbols are present
Organize the images and figures properly
Use subfigure package if allowed by the template
Distribute images over the article, prefer “top” placement
Page-wide placement is often even in two-column style (figure*,
table*)
Use short but self-explaining captions
Many readers inspect only title, abstract and figures
Always reference each figure in the text
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 12 / 27
Charts
A powerful mean of data visualisation
Comparing data (e.g. gain, acceleration, scaling)
Showing an evolution, growth, development
Statistical evaluation
Find your favourite application to create charts
GNUPlot, R, Matlab, Paraview
My basic requirements:
Allows for creating all standard types of charts
Allows for "scripting" the graph and re-generating (my usual pipeline:
create a script, render a graph, modify interactively, regenerate the
script)
Stores the rendered figure in vectorial format (PDF, PS, SVG)
General rules:
Good readability, comprehensibility, contrast (b/w, colors)
Self-explaining short caption (title is not always necessary)
Correct annotation of all axes (physical units, amounts,
descriptions)
Legend if needed
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 13 / 27
Chart Types: Histogram, Pie Chart
Histogram:
Graphical representation of
distribution of data
Show frequency of data (y-axis)
A 3D version exists (with two range
axis): questionable use in papers
(mainly printed)
Pie chart:
Graphical illustration of proportion in
data
Assuming that data sums up 100%
Often criticized: difficult comparison
across different pie charts (often
(mis-)used by media)
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 14 / 27
Chart Types: Bar Chart
Shows comparisons between
categories (discrete range data)
Can be plotted vertically and
horizontally (important when
shortening the paper)
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 15 / 27
Chart Types: Line Chart
Series of data (depicted by markers),
connected (or not) with a line
Typical for time series or for example
scaling
Grid can be used in order to increase
the readability
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 16 / 27
Chart Types: Bar-and-whisker Plot
Depicts groups of data using quartiles
and extrema
Alternatively, whiskers depict ±σ
(std. deviation), percentiles
Outliers can be shown (point markers)
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 17 / 27
Chart Types: Scatter Plot, Combinations
ScatterPlot
Usually used to display variables
under control of experimenter
(measurements)
Efficient tool for showing
correlations (positive, negative, null),
usually combined with best fit line
Plot Combinations
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 18 / 27
Special Charts
Double-axis charts
Logarithmic scale
Shiny 3D charts: usually not
suitable for paper
Cheating with charts (3D
effect, range/domain shift, . . . )
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 19 / 27
Tables
Show large amount of data in small space
Have a good reason for having a table (need to show exact values)
Data in the table must be easy to read and interpret
Font size comparable to that of the text (may be 1 pt smaller)
Minimize the number of lines: mainly use to separate headers (not
each line)
Some experts say the vertical lines should be avoided completely
The content must be self-explaining (with caption or table legend)
Any symbols used in the table should be defined (ideally in the
caption)
Units in physical entities must be shown: e.g. v [ms−1
]
Amount of text inside the table should be minimized
In some situation, a table-dedicated page can appear
Table can be in landscape orientation (requires turning the page
when reading)
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 20 / 27
Bibliography and Citations
Use BIBTEX if possible (i.e., every time)
Serves as a database of papers (possibility to combine with another
tool as Mendeley)
Several styles available, easy to share, combine, archive (plain text)
Correct typesetting of references according to norms
Think of issuing BIBTEX correctly if you change anything in the .bib
file (multiple executions + latexing again)
Editing the .bbl file generated by BIBTEX should be avoided
The conference/journal template/guidelines usually
provide/prescribe the bibliography style:
Paper description style at the end of the paper
Citation label style in the text ([1], (1), (Turing 1936) etc.
Order of citations (alphabetical, w.r.t. the year, appearance in the
text...); depends on the community
https://kuk.muni.cz/vyuka/materialy/
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 21 / 27
Bibliography and Citations
Be careful when creating your BIBTEX database
Correct category: article, inproceedings, proceedings, inbook,
book, manual, misc, unpublished
Correct items (given for each category): authors, title, year, pages,
publisher, volume, issue
Correct spelling of names: diacritics (using TEX commands), names
separated with “and”
Abbreviations of journal titles
Good place to search for some more space
Official list: http://www.ams.org/msnhtml/serials.pdf
Using “et al.” in citations (Chicago Manual of Style,
Cambridge University Press)
Never for less than three authors
Three authors: when repeated (the first occurrence: all names)
Always for more than four authors
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 22 / 27
Acknowledgment, Short Biographies
Acknowledgment is usually required by many funding projects
Sometimes, the phrase used to acknowledge the institution/project
is given
Be careful about the number of acknowledged project/institution
number/code
Individual persons or teams can be acknowledged as well
Not co-authors, but participated somehow (for example
consultation, access to resources, data)
Acknowledged persons should be informed in advance
Short CV is often required in good journals
Two paragraphs summarizing your eduction, affiliation, major
achievement
You can easily find a “template” from a short bio of an
English-speaking author (not much variance)
Small photo is sometimes needed
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 23 / 27
Supplementary Materials
Images, videos or similar material which is:
Submitted in order to provide additional information to the reviewers
Provided by conference/journal for download for readers once the
paper is published
The material is supplementary not essential
It might never be accessed by the reviwer/reader
Should only explicitly illustrate the results clearly and sufficiently
presented in the paper
Should not contain anything that is not presented in the paper
Videos are very often uploaded as a supplementary material
Recorded by camera or screen capture
Equipped with subtitles linking it to the paper (Kdenlive, iMovie etc.)
Usually limited file-size (rather small format, but keep sufficient
resolution and quality)
Be very careful about codecs
The preparation takes usually quite a lot of time: plan ahead (test
tools when you are not under deadline)
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 24 / 27
Numbers, Units, Dashes, Abbreviations
Every time use SI units if possible (Hz, m, s, Pa, N, Ω)
Avoid abusing math-mode for numbers and alphabet units
Wrong: GHz, ms−1
Correct: GHz, ms−1
Discussions about the space between the number and unit: \, \;
siunitx package: \SI{4.5}{GHz} gives 4.5 GHz
Wrong: processor running at 4.3GHz; mean velocity of 5 ms−1
Correct: processor running at 4.3 GHz; mean velocity of 5 ms−1
Make distinction between hyphen and dash:
- (hyphen): compound words: science-fiction
– (2× -, en-dash): ranges: 3–5 days
— (3× -, em-dash): digression in a sentence: "—":
In this case—and only in this case—we propose to. . .
Frequent abbreviations (use newcommand, def in the header
i. e.: “id est”, “that is”: \def\ie{i.\,e.}
e. g.: “exempli gratia”, “for example”: \def\eg{e.\,g.}
w. r. t.: “with respect to”: \def\wrt{w.\,r.\,t.}
Do not repeat “to” after w. r. t.!
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 25 / 27
Using Symbols
Define only those symbols you really need and use later
Use newcommand or def to define symbols in the header:
Extremely easy to change a symbol for any quantity
LATEX source becomes more readable (equations)
Be consistent: e.g. A for matrices, v for vectors, x for scalars etc.
Be careful with Greek letters: ρ vs. , vs. ε (var- prefix)
Use boldsymbol to be sure you have bold Greek symbol (if
needed): λ, α, vs. λ, α,
Avoid double indexing (index of an index): Axα
If dealing with many matrix, vector operations (tensors in general),
consider Einstein notation (usually used in physics)
Automatic summation over indices repeated in the same term
vi = j Aij xj becomes vi = Aij xj
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 26 / 27
Student Presentations
Will be the subject of the last session
Choose any scientific paper you like
Focus on the way it is written
What is good, what is not
Upload into IS before the session
The presentation is informal
It should last up to 10 minutes
Ulman, Peterlík, Obdržálek (FI MU) Writing a Scientific Article II DUVOD, Autumn 2015 27 / 27