Technical writing

● A much more rigorous process

– good English writing is needed but not sufficient
– the key is to convince others about your work
● logic flow, structure, format, style, convention, etc
– well-developed structure and format in a community
● learn by reading from others and oneself and practice
● relatively easier to master with certain attention paid
– writing progresses throughout the (research) work
● not the final stage before the deadline! rush -> sloppy
● writing improves thinking and the work, and also writing
11
 The article
● A technical “formula” (in our community)
– only three layers of abstraction (or “repetition”)
● title -> abstract -> introduction -> main body
● echoed in between and at the end (conclusions)
– a rough “balanced” structure for the main body
● background (and related work, possibly near the end)
● system model or scenarios (may be combined above)
● modeling, design, analysis (depending on approaches)
● evaluation and results (depending on approaches too)
● discussion and further work (to address concerns so far)
12
 The title
● 10 words or so highlighting your work/impact
– very important for indexing and retrieval
– need to reflect your research subject and approach
– highlight your results/impact if possible
– play fancy styles cautiously; read it out loud
● in most cases, title is not a sentence but follows the rules
– common mistakes
● too general/broad/abstract
● too detailed/specific/lengthy
● reuse the same title with oneself or others
13
 The abstract, keywords, etc
● The 100 words or so highlighting your work
– elaborate your title
– highlight the problem, approach and results
– highlight the impact and conclusions
– also used for indexing and retrieval
● and attract the right reviewers (important for a fair review)
● together with keywords and/or index terms/categories
– common mistakes
● yet another introduction section (length limit nowadays)
● over-promise but fail to deliver eventually in the paper
14
 The introduction
● The single most important section in your paper
– what's the problem? get to the point directly fast
– why is it important and challenging? be realistic
– what have others done? briefly now but to the point
– why are they not enough (for the target)? objective
– what's new in this paper? approach & main results
– real contributions: both direct and indirect
– paper structures
– many people decide whether to stop reading now
15
 Set up the stage
● Background
– be more specific about the problem and importance
● Related work
– a zoom-in approach: not a list of “related” papers
– identify/distinguish the most relevant related work
● likely to be compared with quantitatively later
● System model/scenario/preliminaries
– naturally progress from the above: set up the stage
– explain and justify assumptions and simplifications
● need to be addressed again in further discussion 16
 The main stage
● Highly depending on research approaches
– networking research often uses
● measurement, modeling, analysis, simulation, emulation,
prototyping, experimentation approaches, and so on
● in terms of algorithms, analysis and systems skills
● often involving more than one to be convincing enough
– problem formulation is the key!
● unsolvable formulation still leaves unsolvable problems
● not meaningful formulation leads to meaningless results
– then apply the right tools to the right problems
● often need to make changes for the best fit
17
Why is it done better or differently?
● Convincing others is the key
– first, others should be able to understand you
● why good writing is a prerequisite
– second, others should be convinced by you
● a guided tour: problem formulation, solution & evaluation
● it can be repeated, appreciated & used by others
– when showing the results
● do not just dump figures/tables/numbers there
● explain scenarios/parameters/results and extract insights
● reviewers/readers cannot guess what you want to say
18
 More than just a milestone
● Further discussion
– what about those assumptions and simplifications?
● how do they impact? what if they need to be removed?
– also address the concerns arising along the way
● Conclusions and future work
– echo the problem, approach and results
– echo the contributions and impact
– identify future work items for oneself and others
– research is a pipeline competing w/ others/yourself!
19
 Common errors
● In technical writing
– people will read whatever I wrote
● often they do not, other than your own thesis advisor
– people will understand me anyway: likely not
– lack of clear logic flow and structure
● the sentence/paragraph/outline reduction test
● you: logic flow -> structure -> outline as the writing guide
● others: can they recover the outline without seeing it?
– lack of consistency and attention to details
● format, style, figures/tables/bibitems/curves/points/etc
20

H: take a careful look at IEEE style manual.

 To summarize
● The bring-home messages
– to be submitted in your homework

21