This page contains a Flash digital edition of a book.
DATA ANALYSIS: INFORMATION MINING


Burrowing into the bran tub


Felix Grant assesses statistical information mining techniques


statistical information mining techniques to the combined database contents. As a direct result of the operation other companies, seen as friendly to the small country’s interests, received repackaged and anonymised material suggesting productive new lines of enquiry. Patents resulting from these have already been filed; others are in the pipeline. The large company that was the target of the operation, significantly, remains unaware of the opportunities that exist within components of its own activities, which have never been brought together. Or so, at least, I am told. I have no way to


Schematic illustration of the ADMIRE data management Integration model


Intelligence organs of a fairly small, but technologically proficient, country recently managed to clone the email and document bases of a fairly large, but unsuspecting, company with an active and diverse research operation. Not the research results bases


themselves; just the email and document archives. The immediate benefits of the theft were obvious, mundane, and irrelevant here. More interesting from a scientific computing viewpoint (if no more morally or legally defensible) were the spinoffs from applying


Keep taking the tablets


David Smith, pharmaceutical solutions architect for SAS UK, comments: ‘Most data mining techniques can be applied to pharmaceutical data, and have great potential for answering a wide variety of questions across all parts of the business. ‘Association analysis, for example, can help enterprises to understand which drug combinations are associated with adverse events, and which associations are strong


8


enough to be investigated and addressed. Segmentation techniques can help identify groups of physicians who are potential investigators in a new trial. ‘Variable selection can be used to answer questions, such as which genetic markers determine whether a patient will respond well to a treatment, or which process parameters predict quality control failure of a production batch.


SCIENTIFIC COMPUTING WORLD AUGUST/SEPTEMBER 2010


‘Mining of linguistic information helps to gather insight into pharmaceutical methods and techniques. For example, text mining allows pharmacists to identify groups of adverse events associated with a particular drug, a so-called syndrome effect. Content categorisation can highlight potential drug safety concerns suggested by internet tweets and blog entries, or point to abstracts that researchers should prioritise.’


www.scientific-computing.com


verify the truth of this story, of course. But the young man who tells it to me, sitting in the café at the National Gallery, his hands continually roving across the keyboard and digitiser pad of a notebook as he talks, has never sold me a lemon yet. And, true or not, it illustrates a truth: that much knowledge is locked away in information stores assembled for one set of reasons and never reexamined in other ways. Less melodramatically, and less dubiously, information openly published on the internet forms a huge field within which to prospect potential information seams – ‘The low user entry barrier of the web has resulted in massive amounts of unstructured and weakly structured data referring to objects, concepts, user interests and communities,’ to quote the Digital Enterprise Research Institute at


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48
Produced with Yudu - www.yudu.com