一、博士论文写作中的方法和思想

痛定思痛。 2022-08-09 05:18 232阅读 0赞

本文翻译自微软网站 [http://research.microsoft.com/en-us/um/people/simonpj/papers/giving-a-talk/giving-a-talk.htm][http_research.microsoft.com_en-us_um_people_simonpj_papers_giving-a-talk_giving-a-talk.htm]

此网站主要谈及了有关“How to write a great research paper?”的话题，本文是Other Resources中的一篇文章，作者是The University of Birmingham的Aaron Sloman教授，全文如下。

# Research Student - Notes on presenting theses #

## A fresh view of doing a PhD 读博 ##

一个很隐喻的定义是，博士生应该，或者至少应该给知识的殿堂添加一块砖，它能在科学或者工程领域占据一席之地。  
有些博士生可能偶然间开启新的研究领域，但是这都是不可预知的。少量的成果就已经足够了，另外很重要的一点是，不能被博士生必须提出大量创新的观点而束缚了手脚。作为一个博士论文审稿人（a PhD Examiner），particularly an External，我经常用这样的问题测试“这个学生现在是我们博士生的一员了吗？”。

## Introduction 引言 ##

我所提的建议主要是为了在研究生计划和正在写论文时，提供一些指导，而不是手把手教你写。这些建议主要是有关学术论文，对于课程学位的学生可能也会有些帮助。  
假设我们所研究的领域是人工智能，认知科学或者计算科学：  
我编写这个文档得到其他不少人的帮助，因为太多的研究生在大修（substantial modification）之后，仍然被审稿人要求重新提交论文。这对每一个人都是一种资源浪费，我们应该尝试让我们的论文第一次就被接收，甚至不会被勒令重写，这样才能提高表达和组织能力。  
当然，不可能万无一失的保证我么你的论文不会被勒令重写：不论这个学生和他导师多么的优秀，审稿人依然是不确定的，依然存在审稿人会要求你进行额外的实质性工作的要求，比如扩大图表的范围，修正证明中的一个缺陷，增强一些符号的普遍性，包括一些相关未被提及的文献的讨论，等等。  
下面提到的大部分观点与论文的本质无关，而是论文的表述。论文的表述看似不重要，但是对其评估能够反映出你的交流能力（博士曾经意味着教师）。而且，让你论文结构更加清晰明了的表述能够避免对论文中的观点的误解以及其他缺点，因此能够帮助提高主旨（the substance）。当然，好的表述能力也能帮助其他人领会到你所做工作的精髓，这也有助于对我们知识和理解能力的提高。  
本文档需要被反复重写，以期达到一个与分标题更加清晰明了和更加富有逻辑的论文结构，但是我没有时间。我非常欢迎有助于本文档改进的评论和建议。

## Some general comments 一般性的评论 ##

1. 你并不是期望于写一篇震动世界的论文，也不是多么的华丽，避免长篇累牍的赘述能够使你的论文更加充实。

2. 论文的目标是能够向审稿人展示，你所做的研究能够发表在一或两个杂志期刊上。在很少的情况下，你能够认为能够作为一个整体发表，因为所有的位联系够过于紧密，以至于无法分开发表。

3. 注：仅仅开发一些有用的软件或者代码程序是远远不够的，但是大多数人都是为商业机构做着这些事。虽然他们能够获得大量的金钱，但是他们没有资格获得博士学位。 一篇博士论文必须在某些方面具有最高级的知识，所以你必须在你的项目或技术中包含对最新知识的分析。这就需要你对你的工作和其他人的工作做一个对比，就像在文件中稍后的章节所描述的那样。当然，这也需要你提供你的对你的项目或技术中的优点和缺点进行一个分析，清晰的明确它的限制性，解释为什么别人失败的地方你成功了。

## Good communication is extremely important 好的沟通非常重要 ##

1. 论文必须表明你能够有效的与别人沟通，不仅仅是与你领域的极少数专家，而是其他期望能够找到有趣答案的其他人。这也就意味着必须避免模糊的术语表达和公式符号（除非第一次表述）。  
6. 切记：任何可能会引起误解的东西都将会引起误解！

2. 这也就意味着你应该使用例子来代替模糊的术语表达或公式符号来证明你的一般观点。例子一般都应该清晰明了，并具有针对性的，最好是剪短和易记的。应该避免混乱的例子：好的例子应该是能够最简单的证明你所陈述的问题和你表述的观点。

3. 如果你使用并不常见的公式，最好能够对复杂的公式做出英文解释。甚至那些广为熟知的公式，也并不是每一个人都懂的，所以一旦有疑问就应该给出解释，但是注意不要解释的过多。

4. 如果你介绍很多并不常见的专业词汇，并不仅仅是你所精通的分领域， 而是整个学科的人，你应该用一个特殊的章节对交叉引用术语的地方总结一下。

5. 你应该尝试认为你的读者都是很聪明的，但是却故意曲解你的论文。尤其尝试预测这些主要的能够引起误解的曲解形式，这样能够使用有说服力的解释来对抗他们。

## Motivate the reader 激励读者 ##

1. 交流最重要的就是能够吸引读者，让其感兴趣。所以在论文的第一章节或者章节连接处，都应该确保说明你所做的是多么的有趣或重要。

2. 激励读者的第二个方面就是确保读者明白你所做的是什么，和你为什么这么做。好的说明性的例子也是非常重要的。但是仅仅举例是不够的，你必须提出说明性的例子和通俗性的描述来陈述你以后的工作。当然，这也就会包含消极的特征：比如，这将明显的说明你的程序实现不了的功能，而面对理论，你又无法解释，这反而会引导你接下来要做的工作。

## Structure the thesis 论文的结构 ##

1. 好的交流并不仅仅意味着写作风格，你所做的工作必须有一个好的结构，这样很多重要的观点，比如定义、例子、理论和证明等等，在读者重新阅读论文时，就会很容易的找到他们所想找的东西。所以确保在你交叉引用时，最好能够编码或做好标签。  
2. 所有的章节在主表中都应该给出准确的页码。

## Commonly required sections of a thesis 论文中通常需要的章节 ##

在大多数论文中，下面的章节一般都是必不可少的：

(a) Give a high level overview of the solution, then a more detailed overview, then a still more detailed description. I.e. this is the method of "progressive deepening".  
(b) Present your solution in stages, e.g. describing different techniques or partial solutions separately, followed by a chapter showing how they are combined.  
Even if you adopt (b) it is probably a good idea to have an element of (a) - i.e. start with a high level overview before going into the details.  
These may include detailed mathematical proofs, detailed definitions of formalisms, detailed descriptions of programs or techniques used, and examples of the program's execution if you have developed a program.

### 1.Introduction ###

This should give an overview of the main objectives of the thesis, including an account of why the work is worth doing (motivate the reader!), and a summary of what has and has not been achieved.

### 2.Review of related work ###

it is very important in a thesis that you demonstrate that you are familiar with relevant literature, that you can expound it properly and that you know exactly what your own work adds to the work of others.

### 3.More detailed statement of the problem you have worked on ###

The first chapter need not go in to full technical detail. It should be readable by people who are not experts in your field. A later chapter, which may come before or after the literature review, or be combined with it, can go into full technical detail on the nature of the problem.

### 4.One or more chapters outlining your own solution ###

There are two main strategies that can be followed.

### 5.lustrations/Demonstrations of what the solution achieves ###

This may consist of examples of the execution of the program, solutions to theoretical problems, uses of the new notation you have developed, etc.

### 6.Discussion of possible further developments ###

博士论文是永远不会完全结束的，总会有一些概念、理论、技术或者项目上的限制。Make sure that you have looked for such limitations and that you have some ideas about how further work may reduce them. This could be part of a concluding chapter.

### 7. Discussion of how to evaluate the thesis ###

Show that you know how someone should assess your work. Explain what would count as success or failure for your project. Evaluation of a theory in cognitive science might include doing experiments on people (even if you have not done them you should say which experiments would be relevant.) Evaluation of theoretical work in computer science might include attempting to apply it to design of new languages or hardware, or to software engineering techniques, or simply to the solution of other theoretical problems. Evaluation of a new program or technique would include comparing what it achieves with what is achieved by previously existing software and techniques.

### 8.Conclusion ###

This should summarise the main points of the thesis, evaluate what has been achieved (see discussion of evaluation), summarise the way it compares with prior work, mention limitations and failings, and sketch possible future developments or future research suggested by your work.

### 9.Acknowledgements ###

It is conventional, though not absolutely necessary, to have a section acknowledging the people who have helped, inspired, advised you, the institutions that have supported you etc.

### 10.Bibliography ###

Every item referred to in the thesis must be included here. There are different styles for bibliographies and citations, as described below.

### 11.Appendices ###

These may include detailed mathematical proofs, detailed definitions of formalisms, detailed descriptions of programs or techniques used, and examples of the program's execution if you have developed a program.

## If English is not your native language get help 非英语母语国家怎么办 ##

1.If you are not a native speaker of English you are STRONGLY advised to pay someone who is a native speaker to work carefully through your thesis improving the spelling, syntax, etc. where necessary. The examiners have to be convinced that the English is at least up to the standard required for publication in an English language journal. Examiners differ on the importance that they attach to linguistic competence. Some examiners insist that the thesis be written in good English even if you are not a native speaker of English. Others are willing to make allowances for language if the scientific and technical content is good enough. The best advice is to take no chances and assume you are going to get the first kind of examiner.

2.Even if you don't speak English there is no excuse for spelling errors, as there are spelling checkers available on most computers. Make sure that you double-check the spelling of names of other authors.

由于时间关系，前面中文翻译不够细致，基本是边看边翻译，如果有问题，可以到原文链接观看，另外，本文并不是原文的全部，有兴趣完整阅读的同学，戳这里--->[http://www.cs.bham.ac.uk/internal/research\_students/theses.php][http_www.cs.bham.ac.uk_internal_research_students_theses.php]

2015年10月12日 北京

[http_research.microsoft.com_en-us_um_people_simonpj_papers_giving-a-talk_giving-a-talk.htm]: http://research.microsoft.com/en-us/um/people/simonpj/papers/giving-a-talk/giving-a-talk.htm
[http_www.cs.bham.ac.uk_internal_research_students_theses.php]: http://www.cs.bham.ac.uk/internal/research_students/theses.php