Choosing a Dissertation Topic and Literature Review

A good dissertation topic meets three conditions: it is relevant (important for science or society), new (aimed at a problem that is not fully solved) and feasible (there is enough time, data and resources). The topic is usually narrowed step by step from a broad field: for example, from “artificial intelligence” to “image analysis in medicine” and then to a concrete task such as “detecting pneumonia on chest X-ray images with a CNN”. Consulting a supervisor and reading the “future work” sections of recent papers also help to find a topic.

Once the topic is defined, it is turned into a research question, for example: “How much does transfer learning improve CNN accuracy on a small dataset?”. A testable hypothesis, the aim and the objectives of the research follow from the question; the hypothesis must be measurable, that is, it can be confirmed or rejected by an experiment. Then the object of the research (what is studied), its subject (which aspect of it) and the methods (experiment, modeling, survey) are determined.

A literature review is a systematic analysis of existing knowledge on the topic that reveals the research gap. It is not a list of sources but an analysis that compares and synthesizes them. The search is run with precise keywords and logical operators ("transfer learning" AND "medical imaging") in databases such as Google Scholar, Scopus, Web of Science, IEEE Xplore and arXiv. For every source the author, year, method, data, result and limitations are recorded in a table, which makes writing the review much easier.

Sources must be evaluated critically: pay attention to the journal’s peer review system, the author’s reputation, the number of citations, the openness of the methodology and the reproducibility of results. Rely on primary sources (original papers) whenever possible, and use secondary sources (textbooks, reviews) to understand the subject. Programs such as Zotero and Mendeley organize sources and generate the reference list automatically. Using someone’s work without attribution is plagiarism and violates research ethics.

Key takeaways

A good topic is relevant, new and feasible; it is narrowed step by step from a broad field.
A topic becomes a research question, a measurable hypothesis, an aim and objectives.
A literature review compares sources and reveals the research gap.
Searching uses keywords, logical operators and several scientific databases.
Evaluate sources critically, rely on primary ones and cite them correctly.

Practice tasks

Choose a broad topic from your area of interest, narrow it in three steps and write one concrete research question.
In Google Scholar or arXiv build a search query with keywords and the operators AND / OR, pick ten papers from the last five years and record them in a table: author, year, method, data, main result and limitation.
Compare at least three sources from the table, write a one-page review, state the research gap you found in one sentence at the end and format the sources in Zotero.

Review questions

Test yourself: think first, then reveal the answer.

1What three criteria should a dissertation topic meet?Show answer

It should be relevant, new and feasible (in terms of time and resources).

2How does a literature review differ from a list of annotations?Show answer

A review compares and synthesizes sources by theme and shows the research gap, while a list of annotations describes each source separately.

3How do primary sources differ from secondary sources?Show answer

A primary source is the original work in which results were first published (a paper, a patent, a dataset); a secondary source reviews other works (a textbook, a review article).