Cracking codes

In the abstract, LanguageWare doesn't sound very exciting: it's a software program that can find information in unstructured …

In the abstract, LanguageWare doesn't sound very exciting: it's a software program that can find information in unstructured data, a concept that would make many people shrug with indifference, writes Karlin Lillington.

But put it to work, and it can trawl through patient records to pinpoint anomalies that indicate a developing epidemic or a black spot in care at an institution; it can scan criminal files and filter out possible suspects or high-risk cases from among millions; it can run through insurance claims to detect patterns that indicate fraud.

And thousands of people, from students to professional developers to small and large businesses, can figure out their own uses for the powerful program developed by IBM's Dublin software research laboratory, because it is one of many freely available cutting-edge technologies posted on IBM's alphaWorks download website.

IBM describes alphaWorks as a key channel "for reaching an early-adopter audience that is instrumental in shaping the future technology landscape". It allows anyone to get hold of alpha (ie early) code, experiment with it, and perhaps create their own innovations.

READ MORE

"The goal is to showcase emerging technologies coming out of our software and research lab. It helps us to recognise what's hot in the marketplace," says Laura Bennett, a programmer with IBM software solutions and a member of an alphaWorks executive team visiting Dublin this week. "We put stuff out on the site and see what sticks."

Most of the technologies are in the popular areas of online social networking and content definition and analysis - including trendy topics like mashups, wikis, and content tagging - but other segments such as business process automation figure too.

LanguageWare is already drawing interest. IBM hopes this will translate into people or companies wanting to license such alphaWorks technologies, turn them into a product or service, or further develop them as part of an open-source (freely available) software project.

Bennett says alphaWorks began as a small project 12 years ago, when IBM posted online a few program downloads of new technologies developed by its global labs. The company can only commercialise a fraction of its labs' output, but it recognised that many of their good ideas could be put to use by others.

That attitude alone signalled a sea change for a company that for most of its long history has closely guarded all its research and development and could be highly litigious about its intellectual property. Bennett notes that IBM has recently become even more open with the code, which is now available within a more flexible internet protocol framework.

Generally, code can be downloaded and used for three months. At that point, a licence or other arrangement - such as a product collaboration - is needed for its continued availability. Uses could vary from a small- to medium-sized enterprise wanting to implement the code as a feature of one of its own products or services, to an organisation that wants to integrate a program into its internal network.

The latter has been the case for ThinkPlace - IBM alphaWorks code for creating a multifunctional group discussion forum. IBM uses the forum itself on its intranet, but a range of companies have licensed it from alphaWorks for similar use, including BMW and Wachovia.

Another successful alphaWorks program is Many Eyes, which enables users to upload all types of data sets and have them displayed in a range of graphic ways: bar graphs, pie charts and so on. Many Eyes can reveal patterns that quite literally aren't visible otherwise.

While some users have fun with Many Eyes - one student uploaded the text of the Harry Potter novels to analyse word content, for example - organisations like the New York Times, Lloyd's, the US Geological Survey and Procter & Gamble license it for more serious use.

Bennett says more than 800 programs have been posted on the website, and 40 per cent have gone on to be used as product features or open-source contributions. Last year, some 15,000 companies used code from the website.

As well as hoping that the technologies themselves get used, Bennett says IBM benefits from analysing what gets downloaded, how code gets used, and how users discuss the code in the forum communities built into the alphaWorks website.

"That helps determine commercial interest and impacts development," she says.

After Bennett spoke to UCD computer students this week, computing lecturers and brothers Liam and John Murphy - both also fellows at IBM's Dublin software lab - said they were thinking of ways to get students to use alphaWorks code for projects.

"It's like a library of code, where IBM is saying, 'here are all the books'," says John Murphy.