Patent Analytics Technology
The field of information necessary to conduct effective patent analysis is broad, and therefore patent searches often identify a large corpus of documents. Complete consideration of the corpus of documents is necessary for effective results. Most patent search tools permit only keyword-based queries into the corpus of documents. Using keyword-based search technology, patent professionals submit one or more query terms, often connected by Boolean operators, in an attempt to identify relevant documents within the expansive corpus of patent documents. These keyword-based search tools simply return all documents that contain the search terms and thus do a poor job in identifying truly relevant documents. Instead, discovery of the most relevant documents is left to the patent professional to engage in the time consuming task of reviewing voluminous amounts of information.
Instead of using keyword-based search technology, our proprietary Patent Analytics Technology processes complex queries that specify multidimensional elements for the target search. These complex queries may consist of large amounts of text including claim language, full text patent documents, or even entire patent portfolios. To achieve these results, our Patent Analytics Technology incorporates the available corpus of patent literature, including patent and patent applications from US, international and foreign patent databases.
Using machine-learning techniques, our Patent Analytics Technology processes these complex queries against the available corpus of patent documents to obtain a level of precision and recall not possible with use of keyword-based search technology. To conduct a search, our Patent Analytics Technology performs a quantitative comparison of the complex query to the corpus of patent documents and generates a force directed graph that depicts relevance by distance. In this way, the relative closeness of the complex query to the patent documents is revealed. For example, if the complex query is a portfolio of patent documents, then the Patent Analytics Technology identifies all patent documents relevant to the patents in the portfolio. Relevance is visualized in two-dimensional space through graphical depiction of the documents, as nodes, with relative positioning of the nodes to depict relevance.
By mapping the relative closeness between the complex query and the corpus of patent documents, clusters of patent documents emerge. Each cluster represents a field of technology or science. With our Patent Analytics Technology, subject matter areas are easily identified for both the patent portfolio of interest as well as the patent documents relevant to the portfolio.
Patent portfolio managers often classify patents in business units or product classifications. Our Patent Analytics Technology forms the basis to organize the patents into useful topics, including business units or product classifications most suitable to our client’s business objectives.
Our Patent Analytics Technology provides an interactive environment to explore patents of interest mapped to the universe of patent documents. The illustration below depicts an example output from our Patent Analytics Technology. Each node corresponds to a patent document, and the color of the node represents ownership of the patent. The distances among the nodes depict relevance for the patent documents. As such, nodes, and the patents they represent, clustered together in close proximity are highly relevant to one another. Conversely, nodes that are displayed further apart are less relevant to one another. The ring of nodes on the display represents patent documents less relevant to the patent documents of the center nodes. Subject matter areas are identified for the clusters of patent documents. Some clusters on the graph are shown blown out to reveal additional details of the relevance relationships within a subject matter area.