‎

1. The Three-pass Approach
2. Doing a Literature Survey
- 2.1. Unfamiliar Area
- 2.2. Insights

I recently delved into S. Keshav's "How to Read a Paper," where I discovered the THREE-PASS approach. This method highlighted the inefficiency of my usual reading strategy: starting at the beginning and laboriously working through to the end. Many people, I've learned, adopt a multi-pass strategy. This initial revelation made me rethink my approach, especially as a newcomer to a field.

1. The Three-pass Approach

1.1. First Pass

Quick scan to get a bird's-eye view of the paper. Usually not longer than 10 mins.

1.1.1. Title, Abstract, Introduction

1.1.2. Headings of all Sections & Sub-sections

Ignore everythings else (figures, words inside each sections, etc.)

1.1.3. Conclusions

1.1.4. References

Mentally ticking off the ones already read before.

1.2. Second Pass

Read each section with greater care, but ignore details such as proofs. Usually not longer than 1 hour. In this pass, trying to grasp the main thrust of the paper (with supporting evidence). Try to use this way: Try to tell another virtual guy what makes the author write this paper, and why he/she thinks the conclusion is correct.

There may be some sections that you do not really understand, and there can be few reasons:

1.2.1. 1. The subject matter/field is new to you

1.2.2. 2. Unfamiliar terminology and acronyms

1.2.3. 3. The paper maybe poorly written with unsubstantiated assertions and numberous forward reference

1.2.4. 4. It's late at night, you are tired

The potencial solutions might be:

1.2.5. If it is late, and you don't feel good, just go to sleep (It really works!)

1.2.6. Persevere those questions and go on to the third pass (If there are many)

If you are a starter to this field, maybe better do not jump to third pass, and make notes on what you've get in the first two passes and mark this paper finshed. The reason is, for starters, there usually many questions, some of them are crucial, which are really good questions, but some of them are really not that important (Maybe just because the author forget mention some key points/assuming readers know) and not worth to spend a lot of time to figure out. As a starter, we are not good at distinguish between those points, and spending a lot of time stuck on those less important questions isn't a wise choice.

1.2.7. Return to the paper later, after reading some background material

1.3. Third Pass

Only in those two cases, you may need a third phase:

1.3.1. 1. You are a reviewer

1.3.2. 2. This is a well-knowed and classic paper in its field, and you want to fully understand it.

The key to the third pass is to attempt to virtually re-implement the paper: That is, making the same assumptions as the authors, re-create the work. By comparing this re-creation with the actual paper, you can easily identify not only a paper’s innovations, but also its hidden failings and assumptions.

This pass can take about four or ﬁve hours for beginners, and about an hour for an experienced reader. At the end of this pass, you should be able to reconstruct the entire structure of the paper from memory, as well as be able to identify its strong and weak points.

2. Doing a Literature Survey

2.1. Unfamiliar Area

Paper reading skills are put to the test in doing a literature survey. This will require you to read tens of papers, perhaps in an unfamiliar ﬁeld. How can we apply The Three-pass Approach to help?

2.1.1. 1. Do one pass on three to five papers in the area

Use an academic search engine such as Google Scholar or CiteSeer and some well-chosen keywords to find three to five recent papers in the area. Do one pass on each paper to get a sense of the work, then read their related work sections you will ﬁnd a thumbnail summary of the recent work.

1). Find a recent survey paper:

If you can ﬁnd such a survey, you are done. Read the survey (3-pass), congratulating yourself on your good luck.
2). Otherwise, jump to the second step.

2.1.2. 2. Find the best researchers in this area

Find shared citations and repeated author names in the bibliography. These are the key papers and researchers in that area. Then go to the websites of the key researchers and see where they’ve published recently, That will help you identify the top conferences in that field.

2.1.3. 3. Two passes through high-quality papers

Go to the website for these top conferences and look through their recent proceedings. A quick scan will usually identify recent high-quality related work. These papers, along with the ones that published by those key researchers in step 2, constitute the ﬁrst version of your survey. Make two passes through these papers. If they all cite a key paper that you did not ﬁnd earlier, obtain and read it, iterating as necessary.

2.2. Insights

Here are some of my current insights:

2.2.1. Target Audience of Academic Papers:

Typically, authors write with their peers in mind – professors or PhDs in their field – not beginners (unlike tutorials). Therefore, they might gloss over details they assume are common knowledge. This often leaves gaps for newcomers, which are crucial for understanding the broader message.

2.2.2. Non-linear Composition of Papers:

Authors don't necessarily write papers linearly, from start to finish. Each section often stands on its own, making it unnecessary to read everything preceding it to understand a particular part. For a beginner like me, it's more effective to read selectively, starting with the title, headings, and conclusions, then jumping to areas of interest.

2.2.3. Quality and Clarity of Papers:

Not all papers are well-written or accurate. As a reader, adopting a critical mindset is more beneficial than trying to comprehend every line, which is not only time-consuming but might also be futile.

Table of Contents