SemNA

Semantic Network Analysis

Networks of meaning, made legible.

A desktop app that turns a body of text into a semantic network — so the structure of meaning in your corpus becomes something you can see, measure, and compare. No code required.

Free · runs entirely on your machine · your text never leaves your computer.

What you can do

Two kinds of network

Build explicit co-occurrence networks from word usage, or latent networks from word-embedding similarity — the two ways meaning shows up in text.

Measure the structure

Centrality, brokerage, ego density, community structure, and more — each interpreted for what it means in a network of concepts.

Explore & export

Inspect focal concepts and their neighborhoods interactively, then export a self-contained interactive report or the raw measures.

Before you install

Requirements

macOS: Apple Silicon (M1 or later), macOS 12+.

Windows: 64-bit, Windows 10 or later.

First launch

The app is not yet code-signed, so your OS shows a one-time warning. On macOS: right-click the app → Open → Open. On Windows: More info → Run anyway.

Sample corpora

The same datasets built into the app, packaged as plain-text files. Download them to inspect the inputs, reuse them in your own work, or load a copy yourself — each is a compact, analysis-ready research corpus.

Presidential Inaugural Addresses

Every U.S. inaugural address — a classic corpus for tracking how presidential rhetoric shifts across eras.

60 documents · .txt · 335 KB · Download ↓

FOMC Policy Statements

Federal Open Market Committee statements — the Fed's post-meeting language on monetary policy.

65 documents · .txt · 88 KB · Download ↓

SEC 8-K Executive Changes

8-K filings announcing executive departures and appointments — the language of corporate disclosure.

50 documents · .txt · 73 KB · Download ↓

Embedding models

Latent networks measure meaning by the similarity between word vectors. SemNA ships with GloVe-100 built in, so latent networks work out of the box — no download needed. Want a larger or different pretrained model? Download one in your browser and import it into the app.

Browse embedding models →