Code
# Install remotes if you don't have it
install.packages("remotes")
# Install NetworkIsLifeR from GitHub
remotes::install_github("JPvdP/NetworkIsLifeR")NetworkIsLifeR
Patent analytics, text mining & network science β straight from R
πΈοΈ NetworkIsLifeR
Your all-in-one R toolkit for patent analytics, topic modelling, and network science.
Built for Utrecht Universityβs Innometrics and Data Analytics for Sustainability courses β and useful well beyond the classroom.
v0.0.1 Β· MIT License Β· Early Development
Why NetworkIsLifeR?
Modern innovation research sits at the intersection of three messy worlds: unstructured patent text, large bibliographic datasets, and complex collaboration networks. Most R packages solve one piece of the puzzle. NetworkIsLifeR bundles them all into a coherent, course-tested workflow.
Whether youβre mapping technology landscapes from Lens.org exports, clustering research topics with BERTopic-style embeddings, or visualising co-inventor networks, this package gets you from raw data to insight in far fewer steps.
Core Capabilities
π Patent & Publication Parsing
Read and flatten Lens.org JSONL / JSONL.GZ exports β including nested inventor, assignee, and citation fields β into tidy R data frames ready for analysis.
π§ Topic Modelling
A BERTopic-inspired pipeline: sentence embeddings β UMAP dimensionality reduction β HDBSCAN clustering. No Python environment required for the core flow.
π·οΈ Topic Representation
Label your clusters with human-readable terms using udpipe / quanteda (pure R) or spaCy via the reticulate bridge.
π’ Organisation Cleaning
Fuzzy-match and classify messy company/assignee name strings β a perennial headache in patent data, finally tamed.
πΈοΈ Network Helpers
Lightweight igraph-based utilities for building co-inventor, co-assignee, and citation networks from parsed patent data.
Installation
NetworkIsLifeR is not yet on CRAN. Install the development version directly from GitHub:
Then load it like any other package:
Code
library(NetworkIsLifeR)β οΈ Early Development Notice
This package is at version 0.0.1. The API may change between releases. Pin your dependency to a specific commit in production workflows:
remotes::install_github("JPvdP/NetworkIsLifeR", ref = "abc1234")
A Typical Workflow
The diagram below shows how the packageβs modules connect in a real innovation-analytics pipeline.
Lens.org export (.jsonl.gz)
β
βΌ
parse_lens_jsonl() β patent / publication parsing
β
ββββ tidy patent data frame
β β
β clean_org_names() β organisation normalisation
β β
β build_network() β co-inventor / co-assignee graph
β β
β igraph analysis βββββββββββββββββββββββββββββββΊ Network viz
β
ββββ abstract / claim text
β
embed_texts() β sentence embeddings
β
umap_reduce() β dimensionality reduction
β
hdbscan_cluster() β topic clustering
β
represent_topics() β udpipe / spaCy labels
β
Topic map / wordcloudsQuick-Start Examples
1 Β· Parse a Lens.org Export
Code
library(NetworkIsLifeR)
# Point to your downloaded .jsonl or .jsonl.gz file
patents <- parse_lens_jsonl("my_export.jsonl.gz")
# Returns a tidy tibble with one row per patent
head(patents)2 Β· Clean Organisation Names
Inventor assignee fields in patent data are notoriously noisy ("IBM Corp.", "I.B.M.", "International Business Machines"). The cleaning helper standardises them:
Code
patents_clean <- patents |>
dplyr::mutate(assignee_clean = clean_org_names(assignee))3 Β· Build a Co-Inventor Network
Code
library(igraph)
g <- build_network(patents_clean, type = "co_inventor")
# Basic summary
summary(g)
# Plot with igraph
plot(g,
vertex.size = 5,
vertex.label = NA,
edge.color = "grey70",
main = "Co-Inventor Network")4 Β· BERTopic-Style Topic Modelling
Code
# Step 1 β embed abstracts (uses sentence-transformers via reticulate,
# or a pure-R fallback)
embeddings <- embed_texts(patents_clean$abstract)
# Step 2 β reduce to 2D
reduced <- umap_reduce(embeddings)
# Step 3 β cluster
clusters <- hdbscan_cluster(reduced, min_cluster_size = 10)
# Step 4 β label clusters
topics <- represent_topics(patents_clean$abstract, clusters,
method = "udpipe")
# Inspect top terms per topic
print(topics)Data Included
The package ships with a sample dataset (NW_IPC_NL_regions.csv) containing Dutch regional IPC (International Patent Classification) patent counts β handy for testing network and regional-innovation workflows without needing a Lens.org account.
Code
# Load the sample IPC data
ipc <- read.csv(system.file("NW_IPC_NL_regions.csv",
package = "NetworkIsLifeR"))
head(ipc)Dependencies at a Glance
| Purpose | Package(s) |
|---|---|
| Data wrangling | dplyr, tidyr, jsonlite |
| Network analysis | igraph |
| Text / NLP (R-native) | udpipe, quanteda |
| Dimensionality reduction | uwot (UMAP) |
| Clustering | dbscan (HDBSCAN) |
| Python bridge (optional) | reticulate + spaCy |
Who Is This For?
- Students in Utrecht Universityβs Innometrics and Data Analytics for Sustainability courses looking for a single package that covers the full patent-analytics pipeline.
- Innovation researchers who need reproducible, R-native workflows for technology landscape mapping.
- Data scientists exploring bibliometric or scientometric datasets from Lens.org.
Contributing & Feedback
The project is in active early development and contributions are warmly welcomed.
- Fork the repository on GitHub.
- Create a feature branch:
git checkout -b feat/my-feature. - Open a pull request with a clear description of your changes.
Bug reports and feature requests can be filed as GitHub Issues.
Citation
If you use NetworkIsLifeR in your research, please cite:
van der Pol, J. (2024). NetworkIsLifeR: Utilities for patent analytics,
topic modelling and network science. R package version 0.0.1.
https://github.com/JPvdP/NetworkIsLifeRLicense
NetworkIsLifeR is released under the MIT License. See LICENSE.md for details.
Built with β€οΈ at Utrecht University Β· Rendered with Quarto