Case study: Fast-tracking patents

From pain to gain

Extracting information from patents can be a slow and labor-intensive process, owing to the size of the documents involved and the style in which they are written. Often only a small amount of data is required from any particular patent, but finding it can be very difficult. An interactive text mining system like I2E speeds up this process by using domain-specific ontologies, such as genes, diseases and compounds; by extracting numerical information like amounts and concentrations; and by providing the power to limit searches to within certain regions of the patents.

For example, searches can be constrained to find some terms in the Title, other terms in the Claims section, and a particular company in the Agent field. A key advantage of I2E is understanding the context in which target entities are mentioned in addition to uncovering relationships between them.This is of particular interest in journal articles as well as patents.The flexibility of querying provided by the underlying linguistics and the use of regular expressions gives the user huge potential to find this information fast. 

Reviewing patents for drug dosage amounts and extracting contextual metadata

"Once you see what I2E can do, you won't want to go back to wading through irrelevant documents,”
Associate Director, Safety Assessment,Top-10 Global Pharmaceutical Company