Browse Prior Art Database

Publication Date: 2015-Jan-15
Document File: 3 page(s) / 32K

Publishing Venue

The Prior Art Database


Interacting and visualizing large knowledge graphs is an important capability for domain experts in many fields. Due to its size not all the information can be displayed on the screen. Without a notion of importance it is difficult to provide a meaningful visualization and exploration. We base our visualization engine on a fast approximation technique to compute the importance of nodes in large networks and graphs (including knowledge graphs). The attained importance information can be employed to facilitate an interactive exploration of the knowledge for domain experts, e.g. to spot niche products.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 46% of the total text.

Page 01 of 3


    We address exploration based visualization of large knowledge graphs, i.e. the graph describing relations between patents. Due to the massive number of patents it has become increasingly difficult to adequately assess patent claims and identify promising patent sector targets. Alone in the years from 2000 to 2006 the International Patent Classification (IPC) of the European Patent Office (EPO) recorded more than 760'000 patents.

    Modern patent visualization techniques benefit from graphs that are based on patent citations. These graphs use the patents as nodes and introduces an edge between two patents if and only if one patent is citing the other patent. Henceforth, we will denote this graphs as the citation patent graphs or more concisely as the "patent graphs".

    State of the art solutions for facilitating patent assessment and techniques for quantitative visualization of patents make use of such a patent graph. A diverse set of strategies to improve the interpretability of the resulting patent graph has been proposed, such as for example overlay maps that provide an additional layer of information. Visualizing the patents in patent graphs has the following use cases:

1) Reveal connections between patents.

2) Identify key players in patent categories (IPC).

3) Spot niches and/or synergies in coverage of patent categories.

    We propose a method that facilitates a novel interactive exploration of the patent graph visualization: observing the effects of inserting new patents in a patent graph and using patent centrality values as an overlay map. This has various applications, e.g. spotting niches in the patent graph. On top of illustrating the patent graph such that more central patents can immediately be identified, the impact of a new patent can easily be assessed by injecting patents into the graph and observing the shift in the centrality of the patents. This method benefits from techniques of node centrality values' computation.

    The described method will be a first of a kind tool that is able to deliver information about points 1-3 addressed above fast enough, based on a fast iterative method to compute centrality values, to be useable in an interactive fashion. Additionally, using the centrality values as an overlay map is itself a novel way of decorating the visualization of patent graphs. To the best of our knowledge state of the art solutions only allow a static precomputed and predetermined overlay maps


Page 02 of 3

(i.e. technological distance as discussed in [1]). There is no method available to alter the patent graph interactively and get immediate feedback; explorative analysis by adapting the patent graph is tedious with the current tools.

The proposed method works as follows:

    First the node (patent) centrality values are computed. Subsequently the centrality values are used as an overlay map in the visualization of the graph: higher centrality value means a...