Image-based Retrieval Task
This task is similar to the prior art search task, however the prior art patents should be found based on images and text of a query patent. As this is the first time this task is running, a relatively small set of patents have been chosen so as to enable the participants to gain first experience with patents and their images before being overwhelmed with data.
Three IPC sub-classes are chosen for which patent searchers often rely on visual comparison of images in the patents to find relevant prior art. The IPC sub-classes are:
CHARACTERISTIC FEATURES OF FOOTWEAR; PARTS OF FOOTWEAR
DIAGNOSIS; SURGERY; IDENTIFICATION
SEMICONDUCTOR DEVICES; ELECTRIC SOLID STATE DEVICES NOT OTHERWISE PROVIDED FOR
The target data set contains all images for patent documents in these three IPC sub-classes that have an application date previous to 2002 (291566 images grouped by patent document). Note that each patent usually has more than one image.
The queries consist of the text and complete set of images of a number of patents. The aim will be to find the prior art for these patents based on both image and text information, or alternatively image information alone.
CLEF-IP is supported by the
PROMISE Network of Excellence
(co-funded by the 7th Framework Programme of the European Commission)
The image tasks are supported by the IMPEx project
(funded by the Austrian Research Promotion Agency - FFG)