Reading a map
In short: Each circle (node) is an item, such as a publication, author, venue or term. Lines (edges) link related items; thicker usually means a stronger link. Colors show clusters: groups of items more linked to each other than to the rest. Node size shows the measure picked under Node Config → Weightage, for example citation count or betweenness. Click anything to see its details in the side panel.
The parts of a map
| You see | It means |
|---|---|
| Node (circle) | One item. What it is depends on the map type and unit, for example a publication, an author or a term. The overview panel spells it out under Node is. |
| Node size | The node weightage measure. See Node weightage metrics. If every node is the same size, the map is Unweighted. |
| Edge (line) | A link between two items, for example a citation, a collaboration or a co-occurrence. See Edge is in the overview panel. |
| Edge thickness | The edge weightage measure, or the link's strength when no measure is applied. |
| Color | The cluster the node belongs to. Each map type names its clusters differently, for example "Topic", "Research Group" or "School of Thought". See Map types. |
| Position | Strongly linked items are pulled close together, so neighbors usually belong together. The map's orientation has no meaning; you can flip and rotate it. |
| Labels | Names of the nodes. When zoomed out, labels that would overlap are hidden; zoom in to see more. |
The map overview panel
With nothing selected, the right-hand panel shows an overview of the map:
- The map name, its type and its unit.
- Tiles with the number of nodes, edges and clusters, and Avg. edges per node.
- Top nodes, with two tabs: one ranks by the node weightage measure (its tooltip reads, for example, "Highest document frequency") and Connections ranks by number of links.
- View chips when you've hidden part of the map, for example "Top 500 terms" or "Up to 2015".
- Details: Node is, Edge is, Cluster is, Density, Node weight, Edge weight and Created.
- Chat with this map, to ask a research assistant about it.
The left sidebar also shows the node, edge and cluster counts under the map controls. It says "NO EDGES" if nothing is linked, and "N of M labels" when some labels are hidden to avoid overlap.
Node weightage metrics
Node size shows one measure at a time. Change it from Node Config → Weightage, or from Node weight in the overview panel. The options depend on the map type. See Map controls and styling.
| Metric (as labeled in SciNodes) | In plain language | Good for spotting |
|---|---|---|
| Unweighted | Every node is the same size. | A neutral view of structure. |
| Degree (links) / Co-occurrence degree | How many other nodes this one is directly linked to. | Well-connected items. |
| Weighted degree (strength) | The total strength of all its links, so repeated connections count more. | Items with many strong ties. |
| Betweenness centrality | How often the node sits on the shortest route between two other nodes. | Bridges and brokers that connect otherwise separate groups. |
| Closeness centrality | How near the node is, on average, to every other node it can reach. | Central items that are only a few steps from everything. |
| Eigenvector centrality | A node scores high when it is linked to other high-scoring nodes. | Items influential through who they connect to, not just how many. |
| In-degree (citation count) | Citations received from other items on the map. Citation maps only. | Heavily cited items within your corpus. |
| Out-degree | Items on the map that this one cites. Citation maps only. | Review-like items that cite widely. |
| Publications | Number of corpus publications for the author, organization, venue or country. | The most active producers. |
| References | Distinct cited publications a node contributes to the coupling. Bibliographic coupling maps. | Items that cite many of the same publications. |
| Co-citations | How often the node is cited together with others. Co-citation maps. | Core publications of an intellectual base. |
| Document frequency | Number of publications a term appears in. Keyword and term maps. | Widespread topics. |
| Keyword frequency | Total times a term appears, counting repeats. | The most-used vocabulary. |
| TF-IDF | Frequency adjusted so terms found everywhere score lower. | Distinctive terms rather than common ones. |
| Normalized frequency | Keyword frequency scaled from 0 to 1 against the most frequent term. | Comparing terms on one scale. |
How the map is first sized depends on how it was made. A map from a research question uses the measure shown on its card ("nodes sized by …"). Expert Mode pre-selects a natural measure for each type, such as In-degree (citation count) for citation maps or Document frequency for term maps. Some maps open Unweighted, with every node the same size. Pick a measure to bring out differences.
Edge weightage
| Option | Meaning |
|---|---|
| Binary (unweighted) | Every edge has the same weight. |
| Co-occurrences | Number of publications in which the two terms appear together. |
| Association strength | How much more often two terms appear together than chance would predict. 1 means as often as expected; higher means more strongly associated. |
| Temporal gap | The difference in publication years between the two ends of a citation. Citation maps only. |
Clusters
The Composition list in the left sidebar shows every cluster with its color, name, number of nodes and share of the map. Click a cluster in the list to open its panel. You can also select a node and click Highlight Cluster at the bottom of the left sidebar.
The cluster panel shows:
| Item | What it tells you |
|---|---|
| Members | Nodes in the cluster, and their share of the map. |
| Internal edges / External edges | Links that stay inside the cluster, and links that leave it. |
| Cohesion | The share of the cluster's links that stay inside it, for example "120 of 150 collaborations stay inside." (80%). High cohesion means a tight, self-contained group. Low cohesion means the cluster connects widely to others. |
| Density | The share of all possible pairs of nodes in the cluster that are actually linked. 100% means every node links to every other. Large maps are usually sparse, so values below 1% are normal. |
| Rank | The cluster's position by size, for example "#2 of 9 by size". |
| Avg. edges | Average number of links per member, counting only links inside the cluster. |
| Top nodes | The biggest or best-connected members. |
| Connects to | Other clusters it shares the most links with. |
| Strongest edges | Its strongest internal links. |
The map-level Density in the overview panel works the same way for the whole map.
Cluster names are generated automatically a moment after the map appears. To rename one, right-click it in Composition and choose Rename, or click Rename this cluster in the cluster panel. Reset name brings back the generated name.
Node and edge details
Click a node to see its details in the side panel: its type in brackets, its name, lists such as Publications, Co-authors, Cited publications or Cited by, and a Metrics section. Term and keyword nodes also offer a Definition. Click an edge to see what links its two ends, for example "Shared publications" or "Co-citing publications". Click any publication in a list to open its details. For the actions you can take, see Selecting, merging and excluding.
Tips
- Compare the two Top tabs. A node that is big but has few connections is popular but isolated; one with many connections links different areas.
- Look for nodes with high betweenness that sit between two colors. They often mark bridging publications, authors or ideas.
- Use Timeline mode to see how the map grew over time.
FAQ
Why are all my nodes the same size?
The map is Unweighted, or every node has the same value for the chosen measure. Open Node Config → Weightage and pick a measure such as Degree (links).
Are distances on the map meaningful?
Roughly. Strongly linked items are drawn close together, so neighborhoods matter. Exact distances and the direction a cluster sits in do not.
Why don't I see every label?
Labels that would overlap are hidden when zoomed out. Zoom in, or adjust Labeling & length in Node Style.
What's the difference between degree and betweenness?
Degree counts a node's direct links. Betweenness counts how often a node lies on the shortest route between others. A node can have few links but high betweenness if it is the only bridge between two groups.