Which CS professors do research on the Internet

Literature research

Literature Research

The reading list given to you by your supervisor should only be used as a starting point. It is inherent in a scientific elaboration to deal independently with topic-related literature and to carry out your own literature research.

Action

  • Duration
    • In general, it is advisable to set a binding end time before starting the literature search, as any amount of time can usually be invested in it.
    • Seminar paper: approx. 2 weeks
    • Final thesis (BSc / Master): approx. 2 or 4 weeks
  • search
    • Start with the articles you receive from your supervisor.
    • Examine the publication lists of the authors of these articles if you think you can find further important information there.
    • Examine the journals or conferences in which these articles were published. At the ACM, IEEE, or Elsevier there are often separate search masks for each journal or conference that you can use for this purpose.
    • First try to find the highest quality literature possible (first book, then journal, then conference, then workshop, then web). In general, the IEEE and ACM Transaction on Software Engineering are high quality journals.
    • Vary the search terms. Ask your supervisor for alternative search terms.
    • Combine your search terms with "Survey", "Taxonomy", "Overview", "Classification" to find overview articles that facilitate literature research.
    • Use different search engines.
    • Look for articles cited by several of the sources you already have.
    • On Google Scholar, use the "Recent Articles" function or the advanced search to limit the year and also to find newer articles.
    • Give your supervisor early feedback about your research. He / She can help you quickly sort out inferior or irrelevant literature.
    • If you cannot find the PDF file of an article on the WWW, write the author an email and ask him politely.
  • Reading articles
    • Read selectively: First read the abstract, introduction, conclusion. Look at the pictures. Then decide whether to read the whole article. Don't waste time working through irrelevant work. For scientific work it is essential to develop a "filter" against inferior work and to be able to sort it out quickly due to the scope of the available literature.
    • Read purposefully: Approach a text with specific questions and try to answer these questions while reading. Make a note of your questions BEFORE you start reading and examine the article from the perspective of your question. Often you read an article with a certain question in the background very differently than if you read it without thinking. In most cases, you can skip many details of a solution approach and only examine the points that are relevant to the question.
    • Read critically: Question the statements a text makes. Do not immediately believe everything, even if an article appeared at a conference or in a journal. Cross out areas that you do not understand and ask your supervisor or, if necessary, the author about them.
  • Dealing with literature
    • Structure of your own bibliography: Create your own small bibliography for your subject. Only take on work that you consider essential to your area of ​​expertise. You can then use such a bibliography directly when creating your thesis or when writing publications.
    • Editing of texts: You should have learned how to deal with texts (e.g. marking, writing summaries) in school. For scientific work, it is advisable to create short, own summaries (including bullet points) with comments based on experience from important articles. This allows you to persist what you have read and to quickly reconstruct the most important statements of an article even after a long time. Otherwise there is a risk that after a few months you will no longer know what you have read.
    • Topicality: Create a URL list of the most important journals, conferences, etc. in your field and scan their websites about monthly to stay up to date.

Typical mistakes

  • Unreflective searching on the web: A literature search does not mean that you enter the title of the thesis received from the supervisor in Google Scholar and copy the abstracts of the first 3 search hits into your thesis. Firstly, the keywords used are often not meaningful and it can happen that very similar works with just different names already exist. Second, such search engines only sort according to scientific relevance to a limited extent. If only the number of citations is used as an indicator, newer, more relevant works (with even fewer citations) will only appear further back in the search results. Thirdly, such a search does not provide any real knowledge of your subject. Instead, it is necessary to vary the search terms systematically and to search specifically for the relevant conferences / journals / workshops in recent years. Furthermore, you should use different search engines (see below), which deliver very different results. You should also visit the library and familiarize yourself with the basic literature.
  • Quick task: If no related solution approaches can be found straight away in a literature search, you should not become satisfied too quickly and work out your own solution. The statement "There are no related solutions, I'll be the first to solve this problem!" indicates a problem in 99% of all cases. First of all, you should make sure of the relevance of your scientific question. If the problem were indeed relevant and important, it is very unlikely that other researchers have not already tried to solve it. In this case, it is advisable to understand why others have not yet worked on the same topic. In particular, when doing a new literature search, you should abstract from the specific problem and look less focused for at least similar work. Often there is work with similar methods or approaches in other contexts that only relate to a different domain.
  • Insufficient consideration of basic literature: When researching literature exclusively via Internet search engines, relevant standard works (such as books) are usually not or only insufficiently taken into account. Of course, you cannot work through all the books on a subject, but you should still ask your supervisor about the known basic work and search them for answers to your research question. It is quite possible (especially with more theoretical work) that a web search on a topic is unsuccessful because the topic was dealt with before the advent of the WWW and is dealt with in well-known books.
  • Missing citation of similar approaches: If an elaboration does not cite an important literature reference, this is problematic. It demonstrates an unclean scientific approach. It can then happen that work is repeated without generating a knowledge increment from it. The proposed approach may be inferior or incomplete compared to other work. In any case, it shows the author's lack of knowledge of his subject area. This kind of knowledge is expected from a master’s student who has dealt with a specific topic for 6 months.
  • Missing citation of images: If images are taken from scientific articles, sources must be included.
  • Naive depth-first search: A depth first search involves the recursive tracking of all references of a source. Of course, it makes sense to take a closer look at these works, because they were obviously rated as important by the authors of the present source. The problem with depth-first search is that the number of papers to be read can increase exponentially if you don't set any limits. Therefore, you should first read the entire literature that you receive from your supervisor before continuing your search independently.
  • Incorrect or incomplete literature references: It is not enough to reference the URL of a paper. The indication of the title, author, journal / conference / workshop, etc. is essential.
  • Use of inferior sources: When citing sources that are only available on the web, one should be aware of the low level of credibility due to a lack of peer review. A Wikipedia entry can potentially be edited by anyone (also incorrectly or deliberately misleading). A publication in a scientific journal, on the other hand, is subject to strict peer review guidelines and has been examined several times by experts, which fundamentally increases the credibility significantly. So before you cite a website, think about whether the same information can be drawn from scientifically verified sources.
  • Subsequent literature research: A literature search must always be carried out BEFORE starting to work out your own solution, not afterwards. Only when all doubts The possibilities of existing work have been eliminated, and the relevance of the own question is assured, should be started with a separate approach. No serious scientist wastes his time working on problems that have already been solved or are irrelevant. Of course, however, it is also the case that new insights arise when working out your own approach, so that literature should be searched again during or afterwards.
  • Degradation of related work: It often happens in lectures and elaborations that when related work is presented, only their errors are listed and these approaches are presented as inferior. Of course, the speakers want to use this to motivate their own work. However, you should be aware that your own work is ALWAYS full of limiting assumptions and deficits. In this respect, it is more convincing (especially with young speakers) and less challenging for the audience to appreciate the other work and to present one's own work as an improvement in a detail (which it actually is). Make yourself aware of the meaning of the Google Scholar motto "Stand on the shoulders of giants".

Evaluation of literature

  • Categories
    • Book: comprehensive, rarely up-to-date
    • Scientific journal article (journal): topic-specific, up-to-date, very thoroughly reviewed by experts
    • Conference articles / workshop articles (inproceedings): topic-specific, very topical, less thorough assessment, less mature approaches
    • Text on the web (Misc): quick and easy to find, not reviewed, difficult to reference
  • Evaluation criteria
    • Publication medium: Is it a high quality journal, a high quality conference (usually indicated by the fact that it is supported by the ACM or IEEE)? Beware of "new", not established conferences (1st International Conference / Workshop on ...). Is it a printed (better) or purely electronic publication?
    • Number of citations: Search engines like Google Scholar sort literature according to how often it has been cited.
    • Authors: Gather information about the authors. Articles written by doctoral students tend to be less mature than those written by experienced professors. Find out whether the authors have previously published on the same topic and are more experienced with it.
    • Abstract: While reading an abstract, think about it: Is the problem you are working on clearly described? Is the problem being worked on relevant? Is there a reference to related work? Is there a validation of the results in the form of formal evidence or a case study? If, after reading the abstract or the introduction, you are unable to answer these questions negatively or not, this is often an indication of an inferior publication.

Literature search on the Internet