Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Dieses Kapitel geht auf häufige Fehler ein, die Doktoranden bei ihren Forschungsaktionen begehen, und liefert die richtigen Denkweisen, um diese Fallstricke zu überwinden. Sie unterstreicht die Bedeutung realistischer Ziele, effektiven Zeitmanagements und der Auswahl von Forschungsthemen, die mit den eigenen Interessen im Einklang stehen. Der Text betont die Bedeutung des hohen Ziels für die Einreichung von Arbeiten an hochrangigen Veranstaltungsorten und die Vermeidung der Falle, nur auf Publikationen zweiter oder dritter Ebene abzuzielen. Darüber hinaus wird die Bedeutung der Teilnahme an akademischen Konferenzen für die Vernetzung, Zusammenarbeit und die Entdeckung neuer Forschungsmöglichkeiten diskutiert. Das Kapitel schließt mit praktischen Ratschlägen darüber, wie man die ungenutzte Zeit optimal nutzt, und den Vorteilen eines klaren Plans für zukünftige Forschungsaktionen. Durch die Umsetzung dieser Strategien können die Studierenden ihre Forschungsproduktivität steigern und ihre akademischen Ziele erreichen.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
In this chapter, we discuss some common mistakes regarding the research actions that have been taken by new postgraduate students (including us in the early stage of career), which can be categorized into nine types.
In this chapter, we discuss some common mistakes regarding the research actions that have been taken by new postgraduate students (including us in the early stage of career), which can be categorized into nine types. We will also discuss the correct mindsets for the corresponding mistakes.
3.1 Solely Establish Foundations Based on Taking Courses
Once students transit from undergraduate studies to postgraduate studies, they still think that they can only master knowledge based on taking courses (see Fig. 3.1). Of course, taking courses (especially for some fundamental courses, e.g., advanced algorithm courses for some students who work on theory.) is still important. However, each course can be very broad, which covers a lot of topics. As an example, an advanced database course covers topics in relational databases (e.g., SQL, join, and relational algebra), spatial databases (e.g., R-tree), and graph databases (e.g., graph-based systems). Therefore, some students who mainly work on spatial databases can find that the knowledge in relational databases is not very useful/related to their research. Based on this, some productive students tend to master knowledge by reading research papers related to their research instead (see Fig. 3.1).
Anzeige
Some students may have an illusion that a Ph.D. should have mastered a broad range of knowledge. They have this illusion mainly because of some movies/TV series (e.g., Detective Galileo from Japan) or the interpretation from society. As such, these students may aim to master a broad range of knowledge during the postgraduate studies by taking courses. In fact, this mindset is incorrect. Those successful Ph.D. students (or faculty members) normally have a narrow set of knowledge (with very deep understanding). Using the first author of this book as an example, he is dedicated to research topics for developing efficient algorithms in GIS. Until now, he has only worked on “kernel density estimation”, “K-function”, and “line density estimation” in GIS as a principal author. He has nearly zero knowledge for other research topics (e.g., neural networks). However, he received the title of National Science Fund for Excellent Young Scholars (Overseas), which is a prestigious title in China, with the age of 32, and has published nearly 20 research papers as a first author in top-tier database venues (including SIGMOD, VLDB, ICDE, and TKDE). Therefore, in reality, a Ph.D. should be regarded as “narrow but deep” instead of “broad but shallow” (see Fig. 3.2).
Fig. 3.1
Productive students master knowledge by reading research papers instead of taking a lot of courses
Many students (especially for junior students) do not think that it is important to meet deadlines that they have mentioned before. When we ask them why they miss deadlines, they always have an excuse that they are busy with other things (see Fig. 3.3). However, they never try to meet what they have promised before.
For those students, we would like to say that the supervisors are not idiot. It is possible that you can miss the deadline for the first time. But if you keep missing the following deadlines, most of the supervisors can think that these students have no motivation for conducting research and start ignoring these lazy students. Some supervisors can even kick these students out of their research groups. In addition, it is easy for students to develop this kind of bad habit. The main reason is that a university is not the same as a company. There is no fixed working hours and clear tasks for them. The students can stay up late for playing computer games and drinking with their friends and come back to their offices at 5:00pm (without any idea for what to do next). In order to avoid developing this habit, the students need to have good time management. In other words, they need to have the quantitative goal every day in order to meet each deadline. For example, they can set the following reasonable goals.1
(1) Write one to two paragraphs for the draft today.
Anzeige
(2) Implement the proposed method in these three days.
(3) Survey those papers from the area A and summarize them in the draft in the next two days.
By fulfilling the goal every day, it is likely for students to meet those deadlines (provided by their supervisors/other collaborators). Furthermore, students can find that they can develop the confidence for conducting research after they have always fulfilled these quantitative goals. With this good habit, the supervisor will not set any deadline for those students because they can conduct research independently (see Fig. 3.4).
Fig. 3.4
A productive student can set a reasonable goal every day so that the supervisor does not set deadlines for him/her
Many students may think that it is a painful experience for submitting papers to top-tier venues because of these three reasons. First, those papers are likely to be rejected since the acceptance rate of top-tier venues are normally lower than 25% (e.g., the second round of SIGKDD 2025 only has the acceptance rate of 18.4%). Second, those authors are normally from top universities (e.g., UC Berkeley, MIT, Stanford, Technical University of Munich, Oxford, Tsinghua University, Peking University, NUS, KAIST, and University of Tokyo) in the world. Third, many harsh comments are received by the authors for each submission, which are very hard to be addressed. Therefore, they may only aim to submit papers to some second-tier (or even third-tier) venues (see Fig. 3.5), which should be an easy path for them. Indeed, those second-tier venues can let students have the happy life during their postgraduate studies. However, these second-tier venues are hard for helping students increase their research and presentation skills. Here, we consider this analogy. Imagine that you are playing an online game. If you only fight for those monsters with low levels (e.g., level 1 to level 5), it is hard for you to increase your level. In order to significantly increase your level, what you need to do is to kill those monsters with higher levels (e.g., level 10 and level 20) for earning more experience. As such, suppose that you have a lot of second-tier or even third-tier publications. The hiring committee for an assistant professor in a university will seriously doubt whether this student has enough ability to conduct high-quality research in the future. We can guarantee that those prestigious universities have zero chance to hire anyone who has many weak publications.
Fig. 3.5
Students who do not aim high for submitting papers cannot have a good career
Consider those students who only submit papers to top-tier venues (see Fig. 3.6). Indeed, this can be a painful experience because those papers can possibly be rejected for several times. However, those students can learn the high-quality research and presentation skills based on those comments. Moreover, they can significantly increase their ability (by obtaining more experience) during the submission/rejection process. Ultimately, they can have enough ability to independently submit their own research papers and make them accepted to these top-tier venues. With many top-tier publications (i.e., solid research background), it is easy for them to get the offers from prestigious universities (or top research labs/research-based companies).
Fig. 3.6
Students who aim high for submitting papers can have a good career
Figure 3.7 further mentions the additional reason for why we need to aim high for submitting papers. Suppose that we target for the most difficult and the best venues (e.g., cell, nature, and science (CNS)). Even though those papers are rejected (i.e., failure), it is likely that we can still make them published in some top-tier venues. However, if we only target for the second-tier venues, it is likely that the paper will ultimately be published in the third-tier venues (after the rejection). Moreover, like what we mentioned before, targeting for the best venues can help us increase the ability so that it can increase our chance for making the papers accepted for those venues next time. If we only submit papers to some second-tier venues, we will no longer have the chance to get papers accepted in the best venues.
Fig. 3.7
An additional reason for why we need to aim high for submitting papers
3.4 Select a Research Topic That They Are Not Interested in
Nowadays, many students would like to work on those research topics that are very trendy (see Fig. 3.8), e.g., artificial intelligence (AI), computer vision (CV), natural language processing (NLP), deep learning (DL), and large language models (LLM). However, most of these students do not have any background (i.e., prior knowledge) about these topics. Worse still, some of these students may even do not like these topics. The main reason for why they want to work on these topics is that many companies (e.g., Tencent AI, Alibaba, Huawei, Google Brain, and Amazon) may have the higher chance for providing high salary to hire someone with these backgrounds. Here, we need to emphasize that this mindset is completely wrong based on the following reasons.
Fig. 3.8
A research topic may not be job-oriented. But it must be something that you really enjoy for
Reason 1 (It is impossible to determine whether your research topic is still hot when you graduate.): Observe from Fig. 3.9 that a research topic must not be hot forever. Even though a research topic is hot when you start your postgraduate study, it is possible that this topic will not be hot when you graduate. Consider the experience of the first author of this book as an example. When he started his Ph.D. study in 2014, the hot topic in the database community was crowdsourcing (i.e., leveraging a group of people to work on a single task while minimizing the cost). However, this topic was not hot anymore when it was in 2018 (the time when he submitted the Ph.D. thesis). Therefore, suppose that some students only want to work on a hot topic that is not their interest and think that it can help them find a good job. It is possible that they will ultimately be very disappointed as they may find that it is hard for them to publish any paper (because they do not like that topic) and they may not find a good job (as the topic may not be hot anymore).
Fig. 3.9
A research topic is not hot forever (especially for computer science, which is a fast growing field)
Reason 2 (Even though a research area is hot, it does not mean a company needs to hire you.): Suppose that you are lucky and your research topic is still hot when you graduate. It does not mean that it is easy for you to get a job in a company. The main reason is that many students may also work on this topic at the same time (see Fig. 3.10), indicating that there are many competitors for an offer in a company. Even though you work on a research topic that is not hot, it does not mean that you cannot get a job offer. Instead, it is possible for you to easily get an offer because not many people work on this topic, which means you have nearly no competitor for an offer (see Fig. 3.11).
Fig. 3.11
Working on a topic that is not hot does not guarantee that you cannot get an offer of a job
Reason 3 (It is extremely hard for you to produce anything if you do not like it.): When you were young, you had different interests, e.g., playing computer games, playing football, and chatting with friends. At that time, you did these tasks without being told by your mother and your father. The main reason is that you are interested in these tasks. Therefore, you did them even though no one told you to do so. Note that this is the same as conducting research (see Fig. 3.12). If you are interested in one research topic, you will automatically plan for the next tasks and keep having progress. However, if you are not interested in that topic, you may not be motivated to work on it. You may somehow just want to “fulfill” the tasks from your supervisor (see Sect. 2.1). Therefore, you need to keep asking yourself this question. Which research topic are you really interested in? By answering this question, you should expect that you need to be fully dedicated into this research topic for the next four years.
3.5 Do Nothing During the Idle Time
We believe that many (junior) students may have this experience before. When they need to wait for somethings (e.g., the experimental results, the draft which is currently edited by supervisors, and the information from the collaborators), they will not do anything that can push the progress of the paper (see Fig. 3.13). Consider the first author of this book as an example. In the early stage of his postgraduate study (from September 2014 to January 2017), he simply sat in front of the monitor and waited for the experimental results after he had implemented each method. He did not know what he should do next during the idle time in order to further improve the research paper. Therefore, he was not very productive in that time period. Hence, we need to emphasize that those productive students do not waste their idle time (see Fig. 3.13).
Many junior students have no (concrete) plan for what to do ahead (see Fig. 3.14). Without the plan, it is easy for students to get lost every day. When they sit down in front of the monitor and turn on the computer, they will have nothing to start. Therefore, it is easy for these students to surf the internet for doing nothing related to research (e.g., checking the emails, checking the news, and playing the online games). Ultimately, they will find that they have done nothing (or have only a few progress) for each day. The main reason is that students who want to be successful need to set the concrete goals (e.g., finish the kernel density visualization topic before 30th June 2021). With these goals, they can set the plans for achieving them. Once the students have their plans, they can automatically know what they should do when they go to their offices. Here, we emphasize that only those students (with clear plans) can be productive and have potential to be a successful faculty member (see Fig. 3.14).
Fig. 3.14
An unproductive student has no plan for what to do next while a productive student plans for several paper submissions ahead
3.7 Only Want to Find Internships During the Postgraduate Studies
Many students may not want to be fully dedicated to conduct research during their postgraduate studies. Instead, they only want to find internships, which may even not be related to their research, for earning money or working experience. All they want to get is simply a postgraduate degree certificate so that many companies are willing to recruit them with high salaries (see Fig. 3.15). For those students, we would like to emphasize the following three points.
Fig. 3.15
Some students only want to find internships during the postgraduate studies
Those students do not need this degree. Note that a postgraduate degree (excluding a taught postgraduate degree) is mainly for students to conduct research, write research papers, and present research work. These students will be trained in a way that they can have enough ability to work in a university, a research lab, and a research company. Suppose that those students only want to find internships that are not related to their research. It is not meaningful for them to get this degree for staying in academia. Instead, they can find a full-time job, which has a much higher salary compared with the one of an internship position. Some students may have the false hope that this certificate can allow them to get a decent job in a top research lab (e.g., Tencent AI and Microsoft Research). However, these research labs only recruit those students who have solid research outcomes. With this bad attitude, we believe that these students may not have enough top-tier research publications for those labs to consider.
Those students will be very painful during the postgraduate studies. Since those students may not be willing to conduct research, they will not enjoy what they work on (e.g., reading papers, writing drafts, and thinking new solutions) every day. They may find some excuses to avoid the tasks assigned by their supervisors or may simply “fulfill” the tasks by their supervisors (see Sect. 2.1). Therefore, their supervisors may feel extremely angry about their progress and regularly blame them. Furthermore, once the graduation date is close, they will be anxious about their theses because they have not done anything.
The supervisors will not let those students graduate if they do not have enough research outcomes. Many supervisors normally have their requirements for graduation. Using Ph.D. students as an example, they normally need to finish three research topics in four years in order to write a Ph.D. thesis for graduation. Suppose that the students have no faith to conduct research. They may not be able to write one research paper, not to mention three research papers. As such, these students may ultimately need to defer their graduations, causing anxiety and losing the precious time for earning money and working experience.
3.8 Stick to the Same Problem Setting as a Previous Paper
Many postgraduate students may think that their new research papers must follow exactly the same problem setting as a previous paper in order to make sure that they do not solve the “wrong” problem (i.e., do the “real” stuff). They have this mindset mainly because of the following reasons.
They are afraid to motivate a new problem setting. Some junior students may not have enough writing skills/background knowledge. Therefore, they are afraid to write the abstract/introduction for discussing the background of the problem setting (see Fig. 3.16). Therefore, these students may choose the “safe” route by simply following exactly the same problem setting as the previous paper. The main reason is that they think that they can somehow “copy” the introduction from these papers so that they do not need to “write” the introduction.
Fig. 3.16
A student can be afraid to motivate a new research problem
They do not understand the background of that problem well enough. Some students may not understand the background of their research problems well enough. Consider the first author of this book as an example. When he was a junior Ph.D. student (roughly from 2014 to 2015), he worked on a fundamental research topic in image processing, called template matching. Note that this operation aims to find the image patch from a large image that is the closest to a given template image (see Fig. 3.17). Due to the efficiency issues mentioned by many image processing/pattern recognition papers, he followed exactly the same problem setting and focused on developing efficient algorithms for this operation. However, the main reason for why he sticked to this problem setting is that he did not understand the background for why users need to adopt this operation. Moreover, he also did not acquire enough writing skills for motivating new settings of this problem.
Fig. 3.17
Illustration of the template matching problem, where the image patch covered by the yellow box is the closest one to the template image. (Obtained from Fig. 1a and Fig. 1b in “Tsz Nam Chan, Man Lung Yiu, Kien A. Hua. Efficient Sub-Window Nearest Neighbor Search on Matrix. IEEE TKDE 2017”)
They think that solving the same problem setting can advance the state of the art. Some junior students may think that publishing a paper is similar to a competition/an examination, which means that the problem setting must be the same and they need to develop the best solution that can achieve the state-of-the-art performance in order to publish a research paper.
For the above students, we would like to emphasize that their mindsets may be wrong due to the following reasons.2
Easy to produce results for new problem settings. Note that those research problem settings that have been done by other researchers are hard for students to further develop a new solution for improving the performance. Instead, the new research problem settings can be easy for students to produce results (see Fig. 3.18). The main reason is that other researchers may have already developed strong solutions, which can be much better than the naïve one. Therefore, it is quite hard to develop another solution that can be much better than the existing solutions, i.e., only a small room (or even no room) for improvement (see Fig. 3.19).
Fig. 3.18
It is relatively easy for students to produce results for a new problem setting
Relatively hard to make papers accepted with the same problem setting. Even though a student can manage to submit the paper with the same problem setting as the previous paper, it also has the high chance for the reviewer who published that previous paper to review the paper from the student. Normally, if the reviewer is very familiar with that topic, he/she can possibly be very harsh to the paper. The main reason is that the reviewer can easily detect any mistakes/flaws (even for very tiny ones) in the paper. Moreover, as the reviewer is the real expert for that research problem, he/she can, unfortunately, have the very high expectation for the methodology part. Therefore, he/she can easily reject the paper based on the technical novelty/contribution issues (see Fig. 3.20).
Fig. 3.20
The submitted paper can be easily assigned to and rejected by the reviewer who published the paper with the same problem setting before
Many students may think that conducting research is to stay in the office (or in the library), think of big research ideas, and do not need to attend any social activities (e.g., academic conferences). They have this mindset mainly because the society (or some TV shows) may depict that scientists/researchers do not have good social skills (e.g., only stay in a room and think of a lot of ideas). However, this mindset is incorrect, which is based on the following reasons.
Human beings are gregarious. Therefore, all of us should enjoy social activities rather than staying alone for a long time. This is why it is nonsense to say that scientists/researchers, who are also humans, do not love social events. Moreover, it is even more nonsense to say that scientists/researchers should have bad social skills in order to be “qualified” scientists/researchers. Academic conferences should be the way for many researchers from worldwide to connect with each other, which should be regarded as happy and important events for scientists/researchers to attend (but not a waste of time). In addition, there are also three important reasons for why we need to attend academic conferences.
Many new ideas can be obtained by attending conferences. Consider the first author of this book as an example. He attended the SIGMOD 2017 conference, which was held in Chicago in May 2017. At that time, he listened to a paper presentation, called “Scalable Kernel Density Classification via Threshold-Based Pruning”. That paper arouses the interest from him during the conference. Moreover, this presentation let him know that (1) kernel-based statistical models are computationally expensive, (2) improving the efficiency of kernel density estimation model is an important direction, and (3) this direction is very interesting. As such, he immediately started this research topic on September 2017 and published the ICDE 2019 paper related to this topic.
\(\bullet \)Tsz Nam Chan, Man Lung Yiu, Leong Hou U. KARL: Fast Kernel Aggregation Queries. ICDE 2019.
With this paper, he further successfully published a lot of papers regarding how to improve the efficiency of handling kernel-based machine learning models and kernel density visualization. Therefore, without attending this conference, he may take a longer time to (or even may not be able to) figure out this topic, which can slow down his research progress (or even lose a lot of research papers).
Many research collaborations can be established during the academic conferences. Using the first author of this book as an example, the first time for him to meet with his collaborator Ryan Leong Hou U, who is a faculty member in University of Macau, was in the SIGMOD 2017 conference (Until now, they have more than 20 publications in prestigious venues.). At that time, they discussed the paper about how to improve the efficiency of computing Earth Mover’s Distance (the first paper collaborated by them together) and that paper was later accepted in the TKDE journal.
Fig. 3.21
The first author of this book obtains the job opportunity with the title of “Research Assistant Professor” in the Hong Kong Baptist University (HKBU) after attending the WISE 2019 conference
Many job opportunities can possibly be available during the academic conferences. Note that many research-based companies, e.g., Huawei, Tencent, Alibaba, Google, and Microsoft, have their counters for some academic conferences. Therefore, some students can utilize this chance to find some internship positions (or even full-time positions after graduation). Moreover, some senior professors may also want to recruit junior researchers (e.g., senior postgraduate students and postdoctoral researchers) into their universities. Therefore, those junior researchers can also make use of this chance to meet with many professors there in order to find faculty positions. Here, we consider the first author of this book as an example. In the early 2020, he was appointed to be the helper in the WISE 2019 conference (this conference is delayed to the early 2020 due to the serious protest in Hong Kong after June 2019). At that time, he met with a full professor (now a chair professor), called Jianliang Xu, in the Hong Kong Baptist University (HKBU). During the discussion in the conference, Jianliang knew that he would like to find a faculty position. As such, Jianliang sent an email on 1st June 2020 to invite him for joining HKBU as a research assistant professor (see Fig. 3.21). Later, he officially joined HKBU on 1st September 2020 and significantly increased his research productivity for the later three years (from 1st September 2020 to 31st August 2023) in HKBU.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
For those students who are ambitious to work on fundamental problems with rich experience (e.g., developing new algorithms for further reducing the worst-case time complexity to solve the matrix multiplication problem), the following reasons do not apply to them.
