Learn from an academic conference

Sources of learning from academic Conferences: Newsletter of the Chinese Society of Computer Science, 10th issue of the 2015 column

David Alain Grier (David Alan Grier)

Translator: Wang Cao Shilei

The academic conference has some elements that define the direction of our research field. Some of my young colleagues claim that academic conferences are far more valuable to their careers than published journal papers. Although I have objected to this concept during my years as Dean, I now have a chord with it.

I know that in the academic conference, you can meet peers, can communicate with each other in the paper is difficult to talk about ideas. I also admit that the field of computing is developing rapidly, and that the content of traditional journals may lag for two or three years or even four years. Nonetheless, the values that I have deep down in my mind still value subjects that need to write informative papers. I often wonder if people will be able to read our conference materials after 100 years, and the other question that makes me want to get the answer is, how can such a powerful technology be produced in such a chaotic and miscellaneous situation.

Essentially, the main part of the conference is not the paper, but the discussion (track). If left to nature, a specific paper tends to solve more and more detailed problems. This is a common phenomenon that accompanies the mature technology field. In the new field, we are trying to find out the big problem and try to do the research that can affect many researchers. In the more mature areas, we tend to set smaller goals, take a smaller step, and the scope of the research results Exchange is getting smaller.

Some people interpret this process of maturation as the creator of the field is a genius of high IQ, and then the researchers have neither the experience nor the wisdom to form. A careful review of the conference papers reveals that this statement is not realistic. Pioneers in a field are often very influential, but their influence is often simply because they define some of the concepts that make up this field. If you look at the meetings of the the 1950s and 60, you can see that many pioneer researchers are not always able to understand the concepts that are now used to, or to master the mathematical tools for research. The same analysis shows that many of our researchers today have deep insight into their field, and their mathematical or research skills are much higher than many of the early pioneers in the field.

As can be seen from a conference in the late 60, some early workers are not always able to grasp the essence of basic concepts. In a software engineering seminar devoted to the study of a new field at the time, a researcher made a clear analysis of the requirements Specification (specifications enable engineers to define the functionality of a new software system). In the course of his narration, he had an argument with a famous engineer who objected to his claim. The famous engineer (I think it is best not to give his name here) believes that the given program specification may violate the rules of operation, thus requiring the program to produce erroneous results.

The speaker admits that such a thing might happen, but he points out that if the client needs such a process, the resulting system is right for the customer. At this time, the famous engineer got up and left the room, and said the meeting was problematic, stupid and so on.

The well-known experience of an engineer with no name mentioned reminds us that no researcher is consistently right, but it does not explain why the issues involved in conference papers are getting smaller. In fact, this phenomenon is caused by several reasons. First, problems in the mature sector are often more difficult, and researchers are completing an experimental or analytical framework that requires more time, energy and financial resources to learn basic theories and initiate research programs. In this process, they often take a step-by-step, gradual approach to their experience.

However, the development of conference papers may originate from a more important impetus. As the entire technology community matures, academic institutions begin to measure the quality of scientific research by calculating the number of papers. Today's labs, and even individual researchers, aim to publish 8 or 10 or even 20 papers a year. As a long-term reviewer, I see an increase in the number of identical papers. These papers begin with a few paragraphs that usually describe a common experimental framework. They differ only in that each paper has its own unique research hypothesis, as well as a discussion of the hypothesis.

In the past, we might have integrated five or six of these papers into a single, informative article. However, we live in the present, not the past. Therefore, we must adapt to the world. In the current world, the seminar has actually become an essential part of scientific research.

In a conference discussion, it is often expected to see many papers and many different authors commenting on the same point of view. In fact, these papers often do have a common theme, but they show many changes in the subject. You will often find that each paper in the same seminar has its own unique perspective, or unique mathematical representations, or even a unique set of basic concepts. It is these differences that can help us understand the research situation more than the general part of the paper.

At a recent academic meeting, I was involved in a seminar on Web semantics. The participants in the discussion tried to gather the ways in which the different Web sites used the statements and to determine whether the statements they used were consistent. Their work is based on information theory, which I have studied, so that I can keep up with their discussion ideas.

The symbolic system used by the participants is completely different. Although they all use the equivalent or approximate equivalent formula in algebra, the definition of information and the method of measuring distance between words are completely different. In short, no one studies the same problem.

In the end, I learned a lesson from this seminar. In my opinion, all the speakers are dealing with the same problem, but no one is giving a useful solution. The root of the problem is that information theory cannot actually be used to measure semantics. It does not actually form a semantic grouping of words. The word "no" and "yes" have diametrically opposite meanings, but when measured using information obtained from their usage on the Web page, they are very similar.

When I finished reviewing web semantic materials, I wanted to find a paper that would guide some direction in this area. This kind of paper is extremely valuable. They can give a clear set of concepts and ideas that will inspire interesting questions. These papers are often written by people who have deep thinking about the field and are fully aware of the limitations of current research methods. As a result, these authors identify potential sets of concepts that can define the field of research and clearly articulate the problems that can be addressed by using these concepts.

Such papers are often rare and valuable. They need a certain amount of time to write. Regrettably, academic conferences often emphasize speed. Nevertheless, if you have the opportunity to think about all the papers in a seminar, a good meeting will give you advice on how to write such an article.

David Alain Grier (David Alan Grier)

Former chairman of the Computer Association of the Institute of Electrical and Electronic Engineers (IEEE-CS), when Computers were human the author of the book.

Translator: Wang Cao Shilei

Learn from an academic conference