Dimanche 29 Décembre 2024
taille du texte
   
Samedi, 10 Décembre 2011 02:41

IBM Sics Cloud on Drug Patents

Rate this item
(0 Votes)
IBM Sics Cloud on Drug Patents

IBM's IP cloud can help find a chemical needle in the patent haystack. (Photo: Flickr/Pranjal Mahna)

Need to scour the world’s patent databases for a particular chemical compound? There’s a cloud for that.

IBM has released the Strategic IP Insight Platform (SIIP), a cloud-based data analytics system designed to give a shot in the arm to businesses’ intellectual property efforts. The system searches the US Patent and Trademark Office, the European Patent Office and the World Intellectual Property Organization patent databases, as well as abstracts from MEDLINE, the US government’s scientific literature database.

SIIP is a cousin to IBM’s Jeopardy master: the Watson supercomputer. It’s essentially an extremely fast entry-level patent analyst. It can perform natural language searches across millions of patents and science abstracts and it can also look for images of chemical compounds. The system also provides visualization of patent information. IBM says the system extracts molecules and chemical information from patents and abstracts within 24 hours of publication.

SIIP is based on the Unstructured Information Management Architecture (UIMA), the technology that underpins all of IBM’s natural language processing systems, including Watson. UIMA is now an open source project in the hands of the Apache Foundation.

The system was developed for IBM’s internal intellectual property staff. Big Blue then tuned the system for life sciences companies, with a particular focus on drug development. “We focused a lot on life sciences upfront because it’s one of the most difficult areas because of the need to look at complex medical and scientific terms,” said Chris Moore, a vice president of global business analytics and IBM’s Global Business Services.

As part of the SIIP launch, IBM donated a database of all patented chemicals — more than 2.4 million compounds — to the National Institutes of Health. The data was extracted from 4.7 million patents and 11 million abstracts from biomedical journals published between 1976 and 2000. The data was put together in collaboration with AstraZeneca, Bristol-Myers Squibb, DuPont and Pfizer. The database is meant to allow researchers to see relationships among chemical compounds to aid cancer research and drug development.

IBM is looking to broaden SIIP’s reach by adding journal publications and content in languages other than English. IBM is also planning to broaden SIIP’s applicability by turning it for the petrochemical, electronics and consumer products industries, said Moore.

IBM is pitching the system as a way to extract and analyze chemical and biological data in the service of research and development, licensing intellectual property, identifying acquisition targets, collecting intelligence on competitors, landing grant applications and supporting open initiatives. Left unsaid was how handy SIIP is for attacking your competitors in court and fending off patent trolls.

Copyright analysis is not part of the plan for SIIP’s immediate future, according to an IBM spokesperson.

Authors:

French (Fr)English (United Kingdom)

Parmi nos clients

mobileporn