Expert Mode
In short: Expert Mode is the manual way to create a map. Give it a Map Name, pick a Map Type, and click Plot Map. Sensible defaults are applied to everything else. Open the optional sections only when you need to fine-tune: corpus slice, corpus expansion, unit of analysis, thresholds, clustering, layout, normalization and weights. Each section has its own Reset to defaults.
Open Expert Mode
Use any of these:
- With a map open, click into the Map queries bar and click Expert Mode above the suggestions.
- In the Query Bank, click Expert Mode at the top right.
- On the Overview tab, click Plot New Map.
Steps
- Open Expert Mode. The dialog says "Configure and plot a new map without leaving the stage."
- Check the Map Name. It is filled in for you as "Type Map — Project name" and changes with the type until you edit it.
- Choose a Map Type: Citation, Co-citation, Co-authorship, Keyword Co-occurrence, Term Co-occurrence or Bibliographic Coupling. See Map types.
- Optional: click Add description and say what the map is for.
- Optional: open any advanced section below and change settings.
- Click Plot Map. The dialog closes and the map builds. See Building a map.
Click Cancel to close without building. What you typed is kept for the next time you open Expert Mode.
Advanced sections at a glance
| Section | Shown for | What it controls |
|---|---|---|
| Corpus slice | All types | Which publications this one map uses. The corpus is not changed. |
| Expand corpus | All types | Adds publications to the corpus permanently before you build. |
| Type Settings | All types | Unit of analysis, counting method and type-specific options. |
| Thresholds | All types | Minimums and caps on nodes and links. |
| Clustering | All types | How clusters are detected. |
| Layout | All types | How nodes are placed. |
| Normalization | Keyword and Term Co-occurrence | How link strengths are scaled before layout. |
| Weights | All types | What node size and edge thickness measure. |
Corpus slice
Narrows the publications this map is built from, for example "2018 onwards" or reviews only. Use Describe the slice to type a sentence and click Turn into filters (an AI feature that uses credits), or Add a filter yourself. The count line shows "Keeps N of M publications". If it says "No publication matches this slice.", Plot Map stays disabled until you loosen a filter. Details: Corpus slices.
Expand corpus
Adds publications to the corpus for good, unlike a slice. Choose Re-run a previous search (tick the earlier searches to repeat) or New search, set How many to add, and click Re-run for N more or Search and add up to N. Plot Map waits until the run finishes; click Stop to end it early. Expansion uses your weekly credits. Details: Corpus expansion.
Type settings and thresholds
All threshold boxes start empty, which means SciNodes uses its own defaults. Co-authorship is the exception: Min publications starts at 3.
Citation, Co-citation and Bibliographic Coupling
These types show Citation Settings, Co-citation Settings or Bibliographic Coupling Settings, followed by Thresholds.
| Option | Choices and default | When to change it |
|---|---|---|
| Unit of Analysis | Citation and coupling: Publication (default), Venue, Author, Organization, Country. Co-citation: Cited publication (default), Cited venue, Cited author, Cited organization, Cited country. | Pick what each node should be. |
| Counting Method | Whole (default), Fractional, Weighted. Shown for author, organization and country units (co-citation: Cited author only). | Use Fractional when many-author publications would otherwise dominate. |
| Min citations | Citation and co-citation. Empty. | Raise it to keep only well-cited items. |
| Min publications | Citation and coupling. Empty. | Raise it to drop entities with few publications. |
| Min co-citations | Co-citation only. Empty (lowest 1). | Raise it for a sparser, more focused map. |
| Min shared refs | Coupling only. Empty (lowest 1). | Raise it so only strongly coupled items link. |
| Max nodes | Empty. Keeps the most cited (or most strongly coupled) entities. | Lower it for very large corpora or a cleaner picture. |
| Max links | Empty. Keeps the strongest links. | Lower it if the map looks like a hairball. |
These three types need reference data. While it is still being fetched they are greyed out in the Map Type list, with the note "Citation, co-citation and bibliographic coupling need reference data, which is still being fetched." See Empty maps and reference data.
Co-authorship
This type shows Co-authorship Settings, followed by Thresholds.
| Option | Choices and default | When to change it |
|---|---|---|
| Unit of Analysis | Author (default), Organization, Country | Pick people, institutions or countries. |
| Merge name variants | Off. Author unit only. | Turn on when the same person appears as "J. Smith", "John Smith" and "J Smith". |
| Counting Method | Whole (default), Fractional, Weighted | Use Fractional to stop large teams dominating. |
| Author cut-off (per publication) | Empty | Set a number such as 25 to skip consortium publications with huge author lists. |
| Min publications | 3 | Lower it to include occasional authors; raise it to declutter. |
| Max nodes, Max links (co-authorships) | Empty | Cap the map size. |
Keyword and Term Co-occurrence
Both types open Term Co-occurrence Settings. Term Co-occurrence starts with Title and Abstract selected. Keyword Co-occurrence starts with Author Keywords selected.
| Option | Choices and default | When to change it |
|---|---|---|
| Term Sources | Title, Abstract, Author Keywords, Index Keywords, MeSH Terms. Pick at least one. A source no publication has is greyed out. | Add keyword sources for cleaner topics; add Abstract for richer but noisier terms. |
| Max phrase length | Unigrams (1 word), Up to bigrams (2 words) (default), Up to trigrams (3 words). Title and abstract only. | Use trigrams for fields with long technical phrases. |
| Min term length (characters) | 3 | Set 2 to keep two-letter acronyms; raise it to drop short fragments. |
| Apply stemming | Off | Turn on to merge forms like "network" and "networks". Keyword sources are never stemmed. |
| Min occurrences | Empty. Recommended 2 or more with title or abstract. | Raise it to keep only terms used in several publications. |
| Max terms | Empty. Keeps the most frequent terms. | Lower it for a simpler map. |
| Max links (co-occurrences) | Empty. Keeps the strongest links. | Lower it for a sparser map. |
Clustering
| Option | Default | What it does |
|---|---|---|
| Algorithm | Leiden | Leiden: well-connected, high-quality clusters. Louvain: classic and fast. Infomap: based on how information flows. DFS: only separates disconnected parts. |
| Resolution | 1.0 | Higher gives more, smaller clusters. Lower gives fewer, larger ones. |
| Min cluster size | 1 | Smallest number of nodes a cluster may have. |
| Merge small clusters | Off | On: clusters below the minimum join the nearest larger cluster instead of being dropped. |
| Random starts | 10 | Independent runs; the best result is kept. |
| Iterations | 10 | Optimization passes per run. More can improve quality but take longer. |
| Random seed | 0 | Set a fixed number to make results repeatable across builds. |
Layout
| Option | Default | What it does |
|---|---|---|
| Attraction | 2 | How strongly linked nodes pull together. Higher makes tighter clusters. |
| Repulsion | 1 | How strongly all nodes push apart. Higher spreads the map out. |
| Random starts | 1 | Independent layout runs; the best is kept. |
| Max. iterations | 1000 | Optimization passes per run. |
| Initial step size | 1.0 | How far nodes move at first. Larger explores more widely. |
| Step size reduction | 0.75 | How fast movement slows each pass. Lower settles sooner. |
| Step size convergence | 0.001 | Layout stops when movement falls below this. Smaller is finer but slower. |
| Random seed | 0 | Set a fixed number to make layouts repeatable. |
Normalization
Shown for Keyword Co-occurrence and Term Co-occurrence only. The header shows "Default: Association strength".
| Method | What it does |
|---|---|
| Association strength (default) | Compares how often two terms co-occur with what their totals would predict by chance. |
| Cosine | Keeps very common hub terms inside their cluster instead of drifting to the edge. |
| No normalization | Uses raw counts. Dense maps clump toward the center. |
Weights
Node weight sets what node size measures. Edge weight sets what edge thickness measures. Each option shows a one-line description. Expert Mode pre-selects the node weight shown in the table; pick Unweighted if you want every node the same size. You can change both later on the map from Node Config and Edge Config. For plain-language explanations, see Reading a map.
| Map type | Node weight (pre-selected) | Other node options | Edge weight options |
|---|---|---|---|
| Citation | In-degree (citation count) | Unweighted, Out-degree | Binary (unweighted) (default), Temporal gap |
| Co-citation | Co-citations | Unweighted, the structural metrics | Binary (unweighted) |
| Co-authorship | Publications | Unweighted, the structural metrics | Binary (unweighted) |
| Bibliographic Coupling | References | Unweighted, Publications (not on the publication unit), the structural metrics | Binary (unweighted) |
| Keyword and Term Co-occurrence | Document frequency | Unweighted, Co-occurrence degree, Weighted degree (strength), Keyword frequency, TF-IDF, Normalized frequency, Betweenness centrality, Closeness centrality, Eigenvector centrality | Association strength (default), Co-occurrences, Binary (unweighted) |
The structural metrics are Degree (links), Weighted degree (strength), Betweenness centrality, Closeness centrality, Eigenvector centrality.
Duplicate warning
If the project already has a map with exactly the same settings over the same corpus, Expert Mode warns: "This project already has a map with exactly this configuration:" followed by its name. Click Open it to go to that map, or click Plot Map to build another copy anyway.
Troubleshooting
Plot Map is disabled and its tooltip says why:
- "There are no items to plot. Please build the corpus first." The corpus is empty. Add publications first.
- "Enter a map name first." Type a Map Name.
- "No publication matches this slice. Loosen a filter first." Remove or widen a slice filter.
- "Wait for the corpus expansion to finish." An expansion is running for this project.
Other messages:
- "Weight metrics are unavailable right now." The weights list didn't load. Build with defaults, then change weightage on the map later.
- "Nothing in this corpus meets the minimum of N. Lower it in the map's advanced settings and generate again." A minimum threshold removed everything. Lower it and plot again.
FAQ
Do I need to change any advanced settings?
No. The defaults give a sound map for most corpora. Change settings when you have a reason: too many nodes, clusters that are too coarse, or a specific measure you want node size to show.
Can I change settings after the map is built?
Weights, styling, thresholds on what's displayed, and the timeline can all be changed on the map. Type, unit, clustering, layout and normalization are fixed once built. Plot a new map to change them. Your earlier map stays as it is.
What's the difference between Max nodes here and the Nodes threshold on the map?
Max nodes in Expert Mode limits what the map is built from. The Nodes threshold in the map's configuration panel only hides nodes from view and can be changed any time. See Map controls and styling.
Why is Keyword Co-occurrence showing Term Co-occurrence Settings?
Keyword maps are built with the same engine as term maps, with only Author Keywords selected. You can add Index Keywords or MeSH Terms as sources.