1、如何阅读论文作者：S. KeShav，译者：计军平摘要学者们需花费大量时间阅读论文。然而，很少有人传授这项技能，导致初学者浪费了大量时间精力。本文提出了一种高效实用的论文阅读方法“三轮阅读法”。同时，本文也描述了如何采用该方法进行文献综述。1概述学者们出于各种原因阅读论文，比如为了准备一场学术会议或者一堂课，为了紧跟自己所在领域的研究进展，或者为了了解新领域而进行的文献综述。一般而言，一名学者每年会花数百小时来阅读论文。高效阅读论文是一项极其重要但却很少被人传授的技能。因此，初学者不得不在自己的摸索中学习这项技能。结果是他们在此过程中浪费了很多精力，并且常常陷入深深的挫败感之中。多年以来，我一
2、直使用一种简单有效的方法来阅读论文。本文对这种“三轮阅读法”进行了说明，并介绍了该方法在文献综述中的应用。2三轮阅读法该方法的关键点在于分三轮阅读一篇论文，而非仔细地从头看到尾。每一轮阅读都在上一轮的基础上达成特定的目的：第一轮了解论文的大意，第二轮了解论文的主要内容（而非细节），第三轮深入理解论文。S. Keshav, 2007. How to Read a Paper. ACM SIGCOMM Computer Communication Review,37(3): 83-84.David R. Cherition School of Computer Science, Universit
3、y of Waterloo北京大学深圳研究生院环境与能源学院12三轮阅读法22.1第一轮阅读第一轮属于鸟瞰式阅读，快速浏览论文。由此决定是否需要进入后两轮阅读。这一轮使用5至10分钟，包括以下四个步骤：1.仔细阅读题目、摘要及导言；2.阅读章节标题，略过其他内容；3.阅读结论；4.粗略地看一下参考文献，识别出你已经读过的文献。在第一轮的最后，你应该能回答以下五个问题：1.类别：这篇论文属于什么类别？是实证量化分析？还是对现有方法进行改进？亦或是提出了一个新的理论？2.背景：这篇论文与哪些论文有关联？分析的理论基础有哪些？3.正确性：凭经验判断，这篇论文的前提假设是否成立？4.贡献：这篇论文的主
4、要贡献是什么？5.清晰度：这篇论文的文字表述是否清晰？基于上述信息，你可能决定不再阅读这篇文章。原因可能是你对论文的研究内容不感兴趣，或者是由于你对该领域不甚了解以致难以读懂论文。2.2第二轮阅读在第二轮中，更加仔细地阅读论文，但是略过细节（如证明过程）。这有助于你边看边在空白处记下要点或者进行评论。1.仔细阅读论文中的各类图片。坐标轴的标注是否准确？统计结果是否带有误差棒，以表明结论在统计上是显著的？诸如此类的常见错误有助于你判断一篇论文优秀与否。2.标注出与你研究相关但是你尚未阅读过的参考文献（这有助于深入了解该论文的研究背景）。第二轮阅读可能要花上一个小时。经过此轮阅读，你应该能掌握这篇
5、论文的内容了可以用简洁的语言向其他人介绍论文的主要结论及相应的依据。对于你感兴趣的、但不是你研究专长的论文而言，这种程度的理解已经足够了。不过，有时即使完成了第二轮阅读仍然无法读懂一篇论文。原因可能是你对研究主题不熟悉，文章中出现的各种术语及缩写阻碍了你的理解；亦或是你无法理解3进行文献综述3文章的证明或者实验技术；也可能是文章写得很烂，包含了没有事实根据的观点；还可能是深夜读文章时你太累了。这时你有三个选择：（1）把论文放在一边，希望即使不看懂论文也能获得事业上的成功，（2）先把论文放一放，去补充阅读相关的背景材料，之后再回过头来阅读论文，（3）坚持下去，进行第三轮阅读。2.3第三轮阅读为了
6、完全读懂一篇论文（尤其当你是审稿人时），需要进行第三轮阅读。这一轮的关键点是基本重现论文，即采用与论文相同的假设，由你重新推演整个工作。通过比较你的工作与论文的工作，你可以很容易地发现论文的创新点，以及隐含的缺点及假设。这轮阅读关注的是细节。你应当识别出论文的所有假设并加以验证。更进一步地，设想如果你是文章的作者，你会怎样论述文章的观点（idea）。将你的设想与论文的实际论述相比较，能更深入地理解论文的证明以及论证技巧。这一过程将有助于提升你自己的论证水平。在第三轮阅读中，你也应当记下有助于自己未来研究的想法（ideas）。对于初学者，这轮阅读需要四至五小时，而对于经验丰富的学者，仅需一小时左
7、右。这轮阅读过后，应当能凭记忆推想出整篇文章的结构，同时能清楚地说明文章的优缺点。更为重要的是，应当能准确地指出文章隐含的假设，缺少的相关文献，以及实验或分析方法中可能存在的问题。3进行文献综述文献综述是对研究者论文阅读能力的检验。在这过程中，你需要阅读几十篇、上百篇的论文，而且这些论文可能并不属于你所熟悉的领域。你应该阅读哪些论文？下面介绍如何利用三轮阅读法来回答这个问题。第一步，使用Google或CiteSeer等学术搜索引擎以及恰当的关键词找到三至五篇近期发表的论文。对这些论文进行第一轮阅读，了解大致内容，然后重点阅读论文的“相关工作”（或文献综述）章节。通过以上阅读，你将会找到近期相关
8、工作的概述，甚至找到一篇近期发表的综述论文。如果找到了这样的综述论文，那么你的搜索工作就结束了，直接阅读这篇论文即可。如果没找综述论文，那么进入下一步。第二步，找到上述论文参考文献中反复出现的论文以及作者。这些论文及作者是你所在研究领域的关键文献及学者。先把论文下载好，然后去那些学者的个人网页，查看他们近期将论文发表在哪些期刊、哪些学术会议上。这有助于你找到所在领域的顶级期刊及会议，因为最好的学者通常会将成果发表在最好的期刊或会议上。第三步，到上述期刊及会议的网站上查看近期发表的论文。快速浏览论文标题4本文作者的经验4就能找到近期发表的高质量相关论文。这些论文连同你在第二步中找到的论文就是你进
9、行文献综述时需要阅读的第一批论文。对这些论文进行两轮阅读。如果这些论文都引用了一篇你之前未列入上述名单的论文，那么找到并阅读之。如有必要，这一过程可以反复进行下去。4本文作者的经验过去十五年，我一直采用三轮阅读法来阅读会议论文、写审稿意见、做文献综述以及在讨论前快速阅读论文。这种规范的方法使我在把握整体内容前避免陷入无谓的细节中去。这种方法也有助于我估算评阅一组论文所需的时间。更为重要的是，我能根据我的需要以及所拥有的时间来灵活地调整论文阅读的深度。5相关文献如果你是作为审稿人对论文进行评阅，那么你可以阅读Timothy Roscoe关于论文评阅的论文1。如果你计划写一篇科技论文，那么你可以浏
10、览HenningSchulzrinne的网站2以及阅读George Whitesides关于此过程的出色总结3。6请求我会根据读者的反馈不断更新本文。如果读者对本文有任何评论或改进建议，可以给我发邮件，也可以到CCRo网站反馈4。7致谢本文的第一版由我的学生起草，他们是Hossein Falaki，Earl Oliver及SumairUr Rahman。感谢他们的工作。我也从Christophe Doit敏锐的评论以及NicoleKeshav出色的编辑中获益匪浅。本文受以下单位或项目资助：加拿大国家科学与工程委员会，加拿大首席研究员计划，北电网络，微软，因特尔以及斯普林特公司（The Nati
11、onal Science and Engi-neering Council of Canada, the Canada Research Chair Program, Nortel Networks,Microsoft, Intel Corporation, and Sprint Corporation.）。参考文献1 T. Roscoe,“Writing Reviews for Systems Conferences,”http:/people.inf.ethz.ch/troscoe/pubs/review- writing.pdf.参考文献52 H. Schulzrinne,“Writin
12、g Technical Articles,”http:/www.cs.columbia.edu/hgs/etc/writing-style.html.3 G.M. Whitesides,“WhitesidesGroup: Writing a Paper,”http:/www.che.iitm.ac.in/misc/dd/writepaper.pdf.4 ACM SIGCOMM Computer Communication Review Online,http:/www.sigcomm.org/ccr/drupal/.How to Read a PaperS. KeshavDavid R. Ch
13、eriton School of Computer Science, University of WaterlooWaterloo, ON, Canadakeshavuwaterloo.caABSTRACTResearchers spend a great deal of time reading research pa-pers. However, this skill is rarely taught, leading to muchwasted effort. This article outlines a practical and efficientthree-pass method
14、 for reading research papers. I also de-scribe how to use this method to do a literature survey.Categories and Subject Descriptors: A.1 Introductoryand SurveyGeneral Terms: Documentation.Keywords: Paper, Reading, Hints.1. INTRODUCTIONResearchers must read papers for several reasons: to re-view them
15、for a conference or a class, to keep current intheir field, or for a literature survey of a new field. A typi-cal researcher will likely spend hundreds of hours every yearreading papers.Learning to efficiently read a paper is a critical but rarelytaught skill. Beginning graduate students, therefore,
16、 mustlearn on their own using trial and error. Students wastemuch effort in the process and are frequently driven to frus-tration.For many years I have used a simple approach to efficientlyread papers. This paper describes the three-pass approachand its use in doing a literature survey.2. THE THREE-
17、PASS APPROACHThe key idea is that you should read the paper in up tothree passes, instead of starting at the beginning and plow-ing your way to the end. Each pass accomplishes specificgoals and builds upon the previous pass: The first passgives you a general idea about the paper. The second passlets
18、 you grasp the papers content, but not its details. Thethird pass helps you understand the paper in depth.2.1 The first passThe first pass is a quick scan to get a birds-eye view ofthe paper. You can also decide whether you need to do anymore passes. This pass should take about five to ten minutesan
19、d consists of the following steps:1. Carefully read the title, abstract, and introduction2. Read the section and sub-section headings, but ignoreeverything else3. Read the conclusions4. Glance over the references, mentally ticking off theones youve already readAt the end of the first pass, you shoul
20、d be able to answerthe five Cs:1. Category: What type of paper is this? A measure-ment paper? An analysis of an existing system? Adescription of a research prototype?2. Context: Which other papers is it related to? Whichtheoretical bases were used to analyze the problem?3. Correctness: Do the assump
21、tions appear to be valid?4. Contributions: What are the papers main contribu-tions?5. Clarity: Is the paper well written?Using this information, you may choose not to read fur-ther. This could be because the paper doesnt interest you,or you dont know enough about the area to understand thepaper, or
22、that the authors make invalid assumptions. Thefirst pass is adequate for papers that arent in your researcharea, but may someday prove relevant.Incidentally, when you write a paper, you can expect mostreviewers (and readers) to make only one pass over it. Takecare to choose coherent section and sub-
23、section titles andto write concise and comprehensive abstracts. If a reviewercannot understand the gist after one pass, the paper willlikely be rejected; if a reader cannot understand the high-lights of the paper after five minutes, the paper will likelynever be read.2.2 The second passIn the second
24、 pass, read the paper with greater care, butignore details such as proofs. It helps to jot down the keypoints, or to make comments in the margins, as you read.1. Look carefully at the figures, diagrams and other illus-trations in the paper. Pay special attention to graphs.Are the axes properly label
25、ed? Are results shown witherror bars, so that conclusions are statistically sig-nificant? Common mistakes like these will separaterushed, shoddy work from the truly excellent.2. Remember to mark relevant unread references for fur-ther reading (this is a good way to learn more aboutthe background of
26、the paper).ACM SIGCOMM Computer Communication Review 83 Volume 37, Number 3, July 2007http:/doi.acm.org/10.1145/1273445.1273458The second pass should take up to an hour. After thispass, you should be able to grasp the content of the paper.You should be able to summarize the main thrust of the pa-per
27、, with supporting evidence, to someone else. This level ofdetail is appropriate for a paper in which you are interested,but does not lie in your research speciality.Sometimes you wont understand a paper even at the endof the second pass. This may be because the subject matteris new to you, with unfa
28、miliar terminology and acronyms.Or the authors may use a proof or experimental techniquethat you dont understand, so that the bulk of the pa-per is incomprehensible. The paper may be poorly writtenwith unsubstantiated assertions and numerous forward ref-erences. Or it could just be that its late at
29、night and youretired. You can now choose to: (a) set the paper aside, hopingyou dont need to understand the material to be successfulin your career, (b) return to the paper later, perhaps afterreading background material or (c) persevere and go on tothe third pass.2.3 The third passTo fully understa
30、nd a paper, particularly if you are re-viewer, requires a third pass. The key to the third passis to attempt to virtually re-implement the paper: that is,making the same assumptions as the authors, re-create thework. By comparing this re-creation with the actual paper,you can easily identify not onl
31、y a papers innovations, butalso its hidden failings and assumptions.This pass requires great attention to detail. You shouldidentify and challenge every assumption in every statement.Moreover, you should think about how you yourself wouldpresent a particular idea. This comparison of the actualwith t
32、he virtual lends a sharp insight into the proof andpresentation techniques in the paper and you can very likelyadd this to your repertoire of tools. During this pass, youshould also jot down ideas for future work.This pass can take about four or five hours for beginners,and about an hour for an expe
33、rienced reader. At the endof this pass, you should be able to reconstruct the entirestructure of the paper from memory, as well as be able toidentify its strong and weak points. In particular, you shouldbe able to pinpoint implicit assumptions, missing citationsto relevant work, and potential issues
34、 with experimental oranalytical techniques.3. DOING A LITERATURE SURVEYPaper reading skills are put to the test in doing a literaturesurvey. This will require you to read tens of papers, perhapsin an unfamiliar field. What papers should you read? Hereis how you can use the three-pass approach to hel
35、p.First, use an academic search engine such as Google Scholaror CiteSeer and some well-chosen keywords to find three tofive recent papers in the area. Do one pass on each pa-per to get a sense of the work, then read their related worksections. You will find a thumbnail summary of the recentwork, and
36、 perhaps, if you are lucky, a pointer to a recentsurvey paper. If you can find such a survey, you are done.Read the survey, congratulating yourself on your good luck.Otherwise, in the second step, find shared citations andrepeated author names in the bibliography. These are thekey papers and researc
37、hers in that area. Download the keypapers and set them aside. Then go to the websites of thekey researchers and see where theyve published recently.That will help you identify the top conferences in that fieldbecause the best researchers usually publish in the top con-ferences.The third step is to g
38、o to the website for these top con-ferences and look through their recent proceedings. A quickscan will usually identify recent high-quality related work.These papers, along with the ones you set aside earlier, con-stitute the first version of your survey. Make two passesthrough these papers. If the
39、y all cite a key paper that youdid not find earlier, obtain and read it, iterating as neces-sary.4. EXPERIENCEIve used this approach for the last 15 years to read con-ference proceedings, write reviews, do background research,and to quickly review papers before a discussion. This dis-ciplined approa
40、ch prevents me from drowning in the detailsbefore getting a birds-eye-view. It allows me to estimate theamount of time required to review a set of papers. More-over, I can adjust the depth of paper evaluation dependingon my needs and how much time I have.5. RELATED WORKIf you are reading a paper to
41、do a review, you should alsoread Timothy Roscoes paper on “Writing reviews for sys-tems conferences” 1. If youre planning to write a technicalpaper, you should refer both to Henning Schulzrinnes com-prehensive web site 2 and George Whitesidess excellentoverview of the process 3.6. A REQUESTI would l
42、ike to make this a living document, updating itas I receive comments. Please take a moment to email meany comments or suggestions for improvement. You can alsoadd comments at CCRo, the online edition of CCR 4.7. ACKNOWLEDGMENTSThe first version of this document was drafted by my stu-dents: Hossein F
43、alaki, Earl Oliver, and Sumair Ur Rahman.My thanks to them. I also benefited from Christophe Diotsperceptive comments and Nicole Keshavs eagle-eyed copy-editing.This work was supported by grants from the NationalScience and Engineering Council of Canada, the CanadaResearch Chair Program, Nortel Netw
44、orks, Microsoft, IntelCorporation, and Sprint Corporation.8. REFERENCES1 T. Roscoe, “Writing Reviews for SystemsConferences,”http:/people.inf.ethz.ch/troscoe/pubs/review-writing.pdf.2 H. Schulzrinne, “Writing Technical Articles,”http:/www.cs.columbia.edu/ hgs/etc/writing-style.html.3 G.M. Whitesides, “Whitesides Group: Writing aPaper,”http:/www.che.iitm.ac.in/misc/dd/writepaper.pdf.4 ACM SIGCOMM Computer Communication ReviewOnline, http:/www.sigcomm.org/ccr/drupal/.ACM SIGCOMM Computer Communication Review 84 Volume 37, Number 3, July 2007
