Seeing the Forest AND the Trees:
An Introduction to the Patent Projection Graphing Prototype (PPGP)
We present a prototype for an online interactive visualization tool for analyzing the semantic proximities of US patent documents that are related to cancer research and treatments.
Test the prototype patent document visualization system: Interactive Visualization.
This tool allows the user to perform keyword searches and then presents visualizations of sets of relevant patent documents clustered by semantic similarity. Semantic similarity is calculated using a combination of word embeddings obtained using the skip-gram algorithm and the t-SNE dimensionality reduction algorithm.
The user may then select individual points in the cluster to view more detailed patent information. This process allows the user to explore the connections between related patents and see more general trends in the semantic shape of the technological space. It is our hope that this tool may serve as one of the starting points for data analysis leading to future innovative approaches to cancer treatment.
Below is an example visualization resulting from a search for "cell".
Test the prototype system: Interactive Visualization.
View/download the source code: available on Github.
Read our Cancer Moonshot Challenge submission document: Cancer Moonshot Challenge Entry.
Read our submission to the Innovative Applications of Artificial Intelligence (IAAI 2017) conference (under review): IAAI submission.