Skip to main content

Map a new research field

In short: To get oriented in a field you don't know yet, create a project with Suggest publications for me and let SciNodes build its Initial Map, a term co-occurrence map whose colored clusters are the field's main topics. Then ask questions such as "What are the main research themes in this field?" in the Map queries bar, look up unfamiliar terms with Definition, and turn on Timeline to see which topics are new. Each stage below links to a detailed guide.

Before you start​

1. Gather a starting set of publications​

When you don't know a field's search terms yet, let SciNodes plan the searches for you.

  1. Open app.scinodes.com and click + New Project.
  2. Type a Project title of at least 4 characters.
  3. Under Current research focus, pick Ideation or Literature Review.
  4. Choose Suggest publications for me and describe the field in a sentence or two under "What's this project about?".
  5. Click Add Suggested Publications and Create Project. SciNodes shows the searches it plans under "Searching for:" and a preview of the top results, then adds the merged set (up to 800 publications) and opens the project.
  6. Remove clearly off-topic publications later in the Library.

If you already know good search terms, use Discover with SciVerse instead.

Details: Creating projects · Discover and Suggest

2. Read the Initial Map​

SciNodes opens the project on Research Maps and builds a first map called Initial Map automatically. It's a Term Co-occurrence map: terms from titles and abstracts, linked when they appear in the same publication.

  1. Look at the colored groups. Each cluster is a Topic. Larger nodes are more important terms.
  2. Click a cluster to see its top nodes, size, cohesion and density.
  3. Click Chat about this … in the cluster panel to have a research assistant summarize the topic.
  4. Open Composition (the cluster legend) to rename a cluster once you know what it's about.
  5. Click any term node to see the publications behind it.

Details: Map types · Reading a map

3. Ask the questions a newcomer asks​

Type a question into the Map queries bar, click Map this question, pick an option and click Explore with map. These questions from the built-in catalog work well for a new field:

QuestionMap SciNodes usually suggests
"What are the main research themes in this field?"Term Co-occurrence (titles and abstracts)
"Which terms must a newcomer learn first?"Term Co-occurrence or Keyword Co-occurrence
"Which subfields does this corpus actually contain?"Keyword Co-occurrence (Author Keywords)
"Which topics are emerging rather than established?"Term Co-occurrence
"How has the field's topic structure changed over time?"Term Co-occurrence, then Timeline
"Where are the gaps and under-connected topics?"Term Co-occurrence
"Which MeSH descriptors dominate this corpus?"Keyword Co-occurrence (MeSH Terms), for biomedical corpora

To browse more questions, open the Query Bank and pick a category such as Topics & Vocabulary, Trends & Time or Gaps & Ideas, then click Map it.

Term or keyword map? A Term Co-occurrence map works on any corpus because it reads titles and abstracts. A Keyword Co-occurrence map uses the keywords authors (Author Keywords) or indexers (Index Keywords, MeSH Terms) assigned, so it needs publications that carry keywords.

Details: Create a map from a question · Query Bank

4. Learn the vocabulary​

  1. Click a term node on a term or keyword map.
  2. Open Definition in the selection panel.
  3. Read the short definition. A badge shows In this project when the term is ambiguous and SciNodes reads it in the sense your project description suggests, or General for a standard definition.
  4. If your explainer language isn't English, switch between English and your language above the definition.

To make terms easier to read, set Node Config → Weightage to TF-IDF. Nodes are then sized by how distinctive a term is to your corpus, not by how often it appears.

Details: Selecting, merging and excluding · Map controls and styling

5. See which topics are new​

  1. Click the Timeline button (the clock icon) at the bottom of the map panel.
  2. The first time, SciNodes shows "Reading publication years…" while it loads the data.
  3. Drag the year slider. On term maps, terms appear from the year they first occurred and grow with their count up to the selected year.
  4. Terms and clusters that appear only near the end of the slider are your emerging topics.

Details: Timeline mode

6. Clean up and keep what you learn​

  • Merge synonyms: select a term, click Merge, then click the term to fold into it and choose a name. This works on keyword and term co-occurrence maps.
  • Hide generic terms: select a node, click Delete in the selection panel and confirm "Are you sure you want to exclude this data point?". It's removed from the current map view.
  • Map part of the corpus: in Expert Mode, add a corpus slice, for example the last five years.
  • Keep notes: click Write Note on a node or cluster, or RIF it to collect it as evidence in Research in Formation.

Details: Expert Mode · Corpus slices · Research in Formation

Tips​

FAQ​

What is the best map to see the main topics of a field?​

Start with the automatic Initial Map, a Term Co-occurrence map. Its clusters are topics. If your publications carry good keywords, compare it with a Keyword Co-occurrence map of Author Keywords.

How do I find emerging research topics?​

Ask "Which topics are emerging rather than established?" in the Map queries bar, or turn on Timeline on a term map and drag the slider to recent years. Terms that appear late are emerging.

Why are some terms too generic?​

Frequent words dominate when nodes are sized by occurrence. Set Weightage to TF-IDF, or exclude the generic node from the map.

Can I combine two terms that mean the same thing?​

Yes, on keyword and term co-occurrence maps. Select one term, click Merge, click the other term and pick a name for the merged node.