Our Patent programs are developments of the full text reading technology used in our detection programs.
Patent Pipeline makes it possible to put in a full application, set of claims or even a complete patent, and get a ranked listing of the patents in the search set which most closely match it. This makes it suitable across the range of users, from examiners, classifiers and patent searchers.
SpanMatch is a more specialised tool, which identifies pairs of required items without need for boolean construction. It was initially developed to identify elements and ranges within claims for comparison with a query specification in English language patents. but we extended it to work with French, German, Chinese, Japanese and Korean originals to avoid the pitfalls of legacy machine translations. We are also developing its use in other similar fields, such as engineering, chemicals and pharmaceuticals.
Both programs use our proprietary algorithms and are available as APIs. The outputs allow us to demonstrate in our colour-coding house style the ways in which the matching has been found, so the user can verify that the programs are returning what would be expected from a manual operation. That is, there is no 'black box' element. The programs are simply doing, at very high speed and with great accuracy, what a human reader would do if they had sufficient time.