Scientific writing

The skills required for science and writing – are they different?

 "The best science is based on straightforward, logical thinking, and it isn't artistic prose that we expect in [scientific texts] -- we expect clarity." [Writing for Computer Science, J. Zobel]

• do not divagate
• do not pad
• no run-on sentences
• one idea/fact per sentence, one line of thought per paragraph

• Be concise and precise
• be unambiguous (e.g., are back references such as “those” or “it” clear?)
• concrete statements instead of vague descriptions such as “many”, “good”, “quite” etc.

• abbreviations. If used, use them consistently.
• footnotes, because they interrupt the reading flow.
• parentheses. Either it is important to say/read → write it, or it is not important → do not write it.

• facts: present tense
• observations (in own experiments, previous studies): Past tense

• introduce (italics, \emph{…})
• re-use consistently
• use existing terms instead of inventing new, own terms

• vary connecting phrases (many authors overuse “However, …”)
• introductory sentence per section
• Refer to each figure/table. Refer at that point where reader should switch from main text to figure/table.

Be critical on your own text.

“… following elementary steps: create a logical organization, use concise sentences, revise against checklists of possible problems, seek feedback. Like many skills, writing improves through practice and a willingness to accept and learn from criticism.” [Writing for Computer Science, J. Zobel]

“There is no excuse for a report that contains spelling errors.” [Writing for Computer Science, J. Zobel]

• do not invent the wheel a second time
• appreciate previous work
• demonstrate your knowledge of the research area
• provide links to other relevant, interesting, background literature

1. books, book chapters
2. review articles
3. journal articles (peer-reviewed)
4. conference articles (peer-reviewed)

• avoid: PhD theses, posters, personal communication
• if you have several choices: primal publication, most recent publication, most important publication, most elegant publication?

• name-year system: “BLAST (Altschul et al. 1990) and FASTA (Pearson 1990) are based on pairwise alignment.”
• number system: “BLAST [11] and FASTA [17] are based on pairwise alignment.”
• Emphasize others work by stating authors and year of publication: “In 1990, two methods based on pairwise alignment have been introduced: BLAST by Altschul et al. [1] and FASTA by W. Pearson [17].”

• as close to the statement as possible
• directly after naming a method/algorithm/tool
• in case of several statements, after the last

• blabla~\cite{KEY1, KEY2}
• blabla~\cite[Chapter 3]{BOOKKEY}
• \nocite{*} for test usage
• \bibliographystyle{abbrv, alpha, …}

 "Tell them what you're going to tell them, 
  tell them, then tell them what you've told them"

• The opening paragraphs can set the reader's attitude to the whole manuscript, so begin well. [Zobel]

• Not much more than 10% of the complete manuscript (about 0.5–1 page)

• Introduction usually composed of three parts:

• Starting like “In this manuscript, …” indicates missing context.
• Funnel:
  • start broadly, e.g. “In sequence analysis, …”
  • get more and more specific very quickly, e.g. sequence analysis → comparison of sequences → alignments → global vs. local

• Now you can go ahead like “In this manuscript, …”
• as cursorily as possible (to not loose the reader due to too many details)
• as detailed as necessary (to be informative and comprehensible)

• Structuring a manuscript well can be difficult. Communicate the structure/logic to the reader.
• short paragraph (separated from above by \medskip)
• one sentence per chapter
• not too monotonic, connect sentences

• Brief summaries at the start and end of each section
• Sentences linking one section to the next

E.g.: “Together, these results show that the hypothesis holds for linear coefficients. The difficulties presented by non-linear coefficients are considered in the next section.”

“tell them what you've told them” … and more:

• Summary
• Impact
• Open questions / future work

• Variables in italics ($ $).
• Do not use the same variable name for different things.
• Avoid sub- and superscripts.
• Do not begin sentences with variable names.
• Repeat the type of variable when using a variable, but be careful with the placing of the variable. E.g.:

“… a sorted list of numbers sampled from the given set S” versus “… a sorted list S of numbers sampled from given set”

• Not every piece of math is an equation or formula. Maybe it is just a term.

• Definitions, Theorems etc.
  • should be self-contained
  • usually contain an (implicit) “Given…” part followed by a “Then …” part
  • should be put into context / introduced / explained. Do NOT just rephrase a complex formula in words, but explain the meaning.

• Beautify:
  • Take care of the spacing, e.g.: +x (sign) versus + x (addition)
  • Take care of line breaks.
  • Do not use 'x' or \times as multiplication symbol. Use \cdot or just a space.
  • paratheses in different sizes: \left( \sum … \right)
  • x_1 + x_2 + \cdots + x_n
  • x_1, x_2, \ldots, x_n

• Did you write concisely (no excessively long sentences; no redundancy)?
• Are paragraphs used reasonably?
• Did you use some variation in the sentence structure and use of introductory phrases? (Phrasebank)

• Are the title and headings consistent with the content?
• Is the style and wording of headings and captions consistent?
• Do all headings have maximum or minimum capitalization?

• Have all terms been defined?
• Were all new terms introduced in italics (\emph)?
• Has terminology been used consistently?
• Are all variable names in italics (Latex math mode $ $)?
• Are variable names used consistently, not for different things, and not at the beginning of a sentence?

• Is every abbreviation (abbrv.) stated in full when first used (like in this example)? Is any abbrv. introduced more than once? Is any full statement subsequently used unnecessarily?
• Is every abbrv. used less then, say, four times? If so, can it be removed?

• Have bold and italic been used logically?
• Are any words hyphenated in some places but not in others?
• Do the parentheses match? Are they necessary?

• Are the graphs/figures all the same size? Is similar font size used in graphs/figures?
• Are figure sources explicitly cited in the figure caption (Figure reprinted from [x].)?
• Are all figures, tables, and algorithms referenced at an appropriate place in the text?
• Are captions of figures and tables as well as definitions, lemmas, theorems etc. self-contained?

• In the references, has each field been formatted consistently? (Automatically done by Biblatex.) Is capitalization correct, e.g., are acronyms or names capitalized (embrace by {} in your bibfile)? Are journal and conference names abbreviated in the same way? Is the style of author names consistent? (Automatically done by Biblatex.) Has a sufficient and consistent set of fields been provided for each reference of the same type (e.g., “location” for books, “last access” for websites)?

• DO USE A SPELL CHECKER as a very last step before submitting the manuscript.