7. Systematic Searching

There are various ways to conduct an effective search for academic literature and other types of information. The steps suggested in the next few sections partly align with what researchers traditionally do, but they also discuss search techniques employed by more recently developed tools. It is good to keep the following in mind while exploring all of this:

1. Trust the process: A clear search strategy will help you assess and discuss the scope of search results. A proper understanding of how search tools function will help you refine that scope. Willingness to reflect and discuss, and to invest time in exploring search tools and techniques will eventually save you time.
   
   
2. Evaluate search results: Search tools, in particular those with AI-powered functionalities, interpret your input and summarize findings. They can be very helpful, but they may also misrepresent sources, overlook nuance, or cite studies that aren’t peer-reviewed or even real. Whether you're using traditional databases or next-generation platforms, verifying relevance, credibility, and context are a fundamental part of systematic searching.
    
3. Document the process: Keep track of your queries, search tools used, and decisions made. Review what worked, adjust your terms or techniques, and rerun searches as needed. This not only improves your results but also ensures transparency and reproducibility in academic work. For advice on managing your references (retrieved sources), consult 5. Using Information.

To perform an effective search for academic literature and other relevant information, a specified approach is needed. A search strategy can help you define what exactly you are looking for, which makes it easier to actually find it. The steps below will help you create a simple search strategy.

1. Define research question: A good research question is clearly formulated, focused, and relevant. It centres on a specific problem or gap in existing knowledge. For example: "How does medical marijuana compare to opioids in providing pain relief to adults with multiple sclerosis?". Medical research disciplines suggest frameworks like PICO (Population, Intervention, Comparison, Outcome) or SPICE (Setting, Perspective, Intervention, Comparison, Evaluation) to formulate a research question.
2. Determine time span: a logical publication time span depends on the topic and specific question. Some research questions require you to take into account all literature published on the topic, for example when you want to create a mapping of published literature. In other cases it is fair to narrow your time span to only recent publications, for example from the moment a disruptive technology exploded innovations in the field, which can help eliminate irrelevant works. It can also be that you are looking for older information in particular, for example to learn how rocket engines were engineered in the sixties. 
3. Pick languages: although English is the language used primarily in academic publishing, there are situations in which sources in other languages need to be consulted. For example, because a specific country has a strong tradition on your topic of interest. Many academic databases index records whose title and abstract are translated to English to make them more findable. When it comes to grey literature, topics that relate to specific geographical areas might require you to search for documents in other languages. Think of national policies or regional regulations. 
4. Choose types of sources: an easy way to narrow your scope is to list the types of sources you will look for. First, determine if you are looking for academic publications exclusively or if you intend to include grey literature. Then, try to list the specific types of sources you are interested in. Peer-reviewed articles? Conference papers? Technical standards? Patents? Some types of sources might be typical for a discipline; clinical trials in medical sciences, or technical standards in engineering sciences. For more elaborate information on types of sources, consult 2. Types of Information.
5. Select search tools: determine which search tools likely retrieve sources that meet the criteria you defined, and allow you to perform an effective search. You can think of multi-disciplinary database Scopus for a general search on peer-reviewed articles, or Espacenet for a search on patents. Be aware that for some search tools, an institutional subscription is needed. The tools the UT subscribed to can be found here. For more elaborate information on search tools, consult 6. Tools for Searching.

Whether you intend to use ‘keyword-based’ search tools like Scopus and Google Scholar or not, the vocabulary that represents your scope will play an important role in your systematic search. The more accurately you can describe what you are looking for, the more likely it is that you will find it. The steps below will help you create an overview of search terms relevant to you research question.

1. Select key concepts: select the concepts that form the core of your research question. These can consist of one or multiple words. For example: ‘How does medical marijuana compare to opioids in providing pain relief to adults with multiple sclerosis?’ You might consider adding words like ‘compare’ and ‘providing’ as key concepts, in case you want to zoom in on a specific part of the question during your search.
   
   Example of key concept: opioids
2. Define synonyms: for each key concept, list a few synonyms that will help you find sources that use different terms for the same or a very similar phenomenon. For example, synonyms for the key concept ‘opioids’ are ‘painkillers’ and ‘narcotics’. Using Google or an online thesaurus is an effective way to find synonyms. Some databases, like PsychINFO and PubMed contain built-in thesauruses.
   
   Examples of synonyms: painkillers, narcotics, tranquilizers
3. Define narrower terms: for each key concept, list terms that are more specific. For example, if the key concept is ‘medical marijuana’, narrower terms could be the various types of medical marijuana, or the brands that sell them. Using Google or generative AI, you might find narrower terms but make sure to check if you would qualify the suggested terms as ‘narrower’ in the context of your research question. For medical sciences, the hierarchy of MeSH terms in PubMed can reveal narrower terms.
   
   Examples of narrower terms: morphine, heroin, codeine
4. Define broader terms: for each key concept, list terms that are more general. For example, if the key concept is ‘adults’, broader terms could be ‘people’ or ‘humans’. Using Google or generative AI, you might find broader terms but make sure to check if you would qualify the suggested terms as ‘broader’ in the context of your research question. For medical sciences, the hierarchy of MeSH terms in PubMed can reveal broader terms.
   
   Examples of broader terms: drugs, medicine

To perform keyword-based searches in databases like Scopus or search engines like Google Scholar, you need to create queries or search strings. For maximum control over a query, it is best to make use of the ‘advanced search’ option that can be found in most keyword-based search tools. The query has to be formulated in the ‘language’  or search syntax that the particular search tool understands. For example, whereas Scopus uses TITLE-ABS-KEY to indicate a search within the title, abstract and keywords of a record, Web of Science uses TS.

For Scopus, a query could look like this: 

TITLE-ABS-KEY (“medical marijuana” AND (opioids OR painkillers) AND adult*)

Below, the way in which each part of query refines the scope is explained. An overview of these and more ways to refine your search can usually be found in a ‘search tips’ section on the advanced search page. For Scopus, search tips can be found here.  

• Field codes / field tags - TITLE-ABS-KEY: Commands that tell a database to search within specific parts of a record, like title, abstract, or author. Field codes or tags will help you target your search more precisely and avoid irrelevant results.
• Boolean operators - AND, OR, NOT: Logical connectors used to combine or exclude search terms, with the aim of broadening or narrowing the scope of your results. Be aware that through exclusion (operator NOT) you easily unintentionally filter out relevant papers, so use this technique wisely.
• Truncation - adult*: An asterisk added to a word stem to find all its possible endings. Be aware that you risk including unwanted endings, in this case for example adultery.  
• Exact phrasing - “medical marijuana”: Using quotation marks to search for words in a specific order as a fixed phrase. Effective in particular when searching in full-text databases where the individual words could show up in different section of a record on a completely different topic.

Item-based search is a technique that begins with a known, relevant publication—often called a seed article—and uses it as a starting point to discover related literature. Common ways to explore connections are listed below.

Citation relationships can be discovered through reference searching, also referred to as snowballing. Backward reference searching involves examining the reference list of a relevant article to identify earlier works that informed it - essentially looking into the intellectual past. Forward reference searching tracks newer publications that have cited the article since its publication, helping you follow the evolution of ideas. In a multi-disciplinary abstract database like Scopus, each article record is linked to its cited references and citing documents within the database, allowing for easy navigation through related literature.

Shared references can be visualized by generating dynamic citation maps that for example show co-citation or bibliographic coupling. Co-citation occurs when two papers are cited together by a third paper, suggesting they are related in content or influence. Bibliographic coupling links papers that cite the same sources, revealing thematic overlap or shared foundations. Search tools like VOSviewer help researchers explore shared references visually, making it easier to identify influential works, emerging trends, or overlooked studies.

Natural language searching uses free-text queries, often phrased as questions or descriptive statements (e.g., “What are the effects of microplastics on marine life?”). Tools that employ this technique explore  how closely papers align in content, topics, or research questions, regardless of whether they cite each other. This thematic similarity is often determined by analysing keywords and abstracts, full-text content, and semantic meaning using AI or NLP models.

Natural language searching is especially useful for discovering conceptually related papers that may not be connected through citations, ideal for emerging fields or interdisciplinary research. Tools like Elicit and Asta specialize in this kind of search.