Summary of "Academic Standards and Paper Writing"

1. Writing Standards

The Importance of Writing

The Use of Writing Tools:

1. Overleaf Multiplayer Collaboration Online Platform
2. Git and TortoiseGit version control
3. Grammarly grammar detection
4. Typesetting in Latex format
5. Firgure pictures are important

A few notes:

• Avoid large and meaningless graphs
  
• The size in the picture should be clearly seen
  
• Writing should indicate the subject and object, avoid invalid description
  
• The pictures should show more innovation and should not be too similar to others
  
• Present results in a more novel way
  

2. Preparation for writing

The first thing to be clear is the innovation of the paper.
The innovation is very important:

1. new problem new solution

2. These problems can be started from another angle, bypassing some common problems, because of those problems, it is difficult for our method to be as competitive as other advanced methods

3. Existing methods and new solutions

4. These ideas and methods have been improved in terms of speed, precision, and memory

5. Combination of ideas: Combining these ideas can complete a predetermined problem

some suggestions

• Try to find only one problem to solve. If several problems are solved at the same time, the reviewers may not believe it
  
• Combination idea: You must have your own insights
  

3. Thesis structure

3.1 Title

The title should be eye-catching and attract the attention of others. It should be easy to search, because the article is written for others to read, and the number of citations of the paper is still very important. Try not to use based on, because it will appear bad street. The length is best controlled between 40-60 letters.

3.2 Summary

The abstract usually includes three parts: a review of the existing work, a description of the work in this paper, and the experimental results. Just follow this.

3.3 Keywords (optional)

Keywords are an important way to retrieve papers. They are often seen as a supplement to the abstract. Generally, 3-5 keywords are required. Sort the keywords alphabetically.

3.4 Introduction

The introduction should describe the whole story in detail. The most template way, is to adopt the same rhythm as the abstract. Each sentence of the abstract expands into a paragraph in the introduction. The first sentence of each paragraph is called the "topic sentence". The paragraph should have 5–10 sentences. This way, the paragraph should have 50-150 words.

The structure of the introduction:

(1) Literature review

(2) Raise research questions

(3) Why do you need to solve this problem (highlight the importance of this paper)

(4) How to solve this problem (the solution is a theoretical perspective)

(5) Paper structure/framework

(6) Foretelling the conclusion

(7) How important is this research to the broad and narrow research space

(8) Raise expectations from different levels

The key to the introduction:

(1) Screen and organize the content of the introduction (do not paste or paraphrase the entire content of the main text)

(2) Narrow down the topic

(3) Each component transitions naturally

(4) Clearly identify and express research questions, scope, methods, and structure

(5) Language expression is correct and effective

(6) Connect past scholarships with the topic/research questions of the paper, not just review/summary

(7) Make a gap/hook/suspense to create a research space, and then write how to fill up the gap/solve the problem

(8) Define related terms

3.5 Related work (research background)

The list of literature reviews, the correct format of mathematical expressions and symbols. The normal display of special characters and other tasks have corresponding rules, which cannot be taken for granted.

A few notes:
(1) It is not a summary of all relevant literature in this study, nor is it a review of all cited literature in this study;

(2) It is necessary to systematically write down the views on the literature and the connection between the literature, highly link the previous research with this research, and form a comprehensive and systematic knowledge network of previous research;

(3) The literature review of the small paper is part of the introduction, and the literature review of the large paper is a separate chapter after the introduction;

(4) Select documents that are highly relevant/high-quality, high-innovation, and high-value to the research question and conduct detailed discussions, reject general sources (such as reference books, textbooks, simple repetition of other people's research), and give priority to core journal articles;

(5) Describe the differences between previous research and this research, so as to establish a narrow research space;

(6) Emphasize the significance and value of this study by summarizing, comparing, refuting, etc.

3.6 Implementation Details

It is the core part of the thesis

and is written according to the design model.

3.7 Experimental part

The main purpose of the experiment is to answer some questions that the authors and readers care about

Mainly compare:

• Internal comparisons: ablation experiments
  showing performance changes due to parameter changes.
  Demonstrating the advantages of the main scheme compared to its variants.
• External comparison: compared with existing algorithms,
  it is necessary to compare classic schemes, benchmark schemes, and the most advanced schemes.
  If there is too much data, it is more appropriate to use a table than a graph.
  The most important comparison is placed at the end.
  There must be enough text for analysis, Readers cannot be allowed to observe by themselves.
  It is not necessary to beat other schemes on all data sets.
  In addition to analyzing the advantages of your own scheme, you should also analyze its disadvantages

3.8 Conclusion

The conclusion should generally not be too long, 5 sentences is enough. Avoid using the same sentences as in the abstract. The abstract says what we have done, but here we should say what observations and conclusions we have obtained, that is, the conclusion is more specific than the abstract.

4. Writing skills

• Avoid using only formulas, and need to have text descriptions.
  
• Algorithm descriptions should be in the correct logical order
  
  and can be described using pseudocode
  
• Experimental methods need to clarify steps and some details for reviewers to believe
  
• Summary requires objective discussion, don't shy away from methodological flaws
  
• References need to mark key areas as further explanations for text descriptions
  

5. Other skills

5.1 Algorithm Pseudocode

1. Input and output need to be explained;
2. The method (function) name can be written or not, if it is called by other methods, it must be written;
3. Notes need to be written out of the main steps;
4. The length is controlled at 15-30 lines;
5. Mathematical formulas or references to existing mathematical formulas may be used;
6. Unimportant steps can be omitted;

5.2 Images

1. The picture should use a vector diagram, which will not be distorted after being enlarged;
2. The legend can be quite long, some journals require the picture to be self-enclosed, that is, only the explanation of the legend is enough
3. Keep diagram descriptions concise
   
   

5.3 Writing language

1. Write step by step
2. present in an order

Writing language should be precise and concise

Pay attention to the use of some phrases

• In the formal papers of Have't and Don't
  , this abbreviation is incorrect. Use have not, do not, cannot, etc. Use apostrophes only in cases such as authors'.

• And
  It is forbidden to use And at the beginning of a sentence. Use and with caution in the middle of a sentence.

• Use Solve
  with caution. Only when you have solved a certain problem and written the corresponding theorems and proofs can you say solve. Doing deep learning generally cannot solve a certain problem, so you should write handle or address.

• Other
  Use the website to check the frequency of use of words or vocabulary. Generally, words with high frequency of use are selected.