In the mashup above, a user has combined an organizational map of a company with employee pictures, profile information from the company directory, and a database of files that each person can update.
To build these applications, a user selects from lists of widgets and data sources and drags them together onscreen. Dropping a list of store locations onto a map widget makes the system automatically plot those locations on a map. Gerken says that a major design challenge was programming the system so that it could understand what the user likely wants it to do in such a situation. To try to solve that problem, the system tries to recognize similarities in data that might not be tagged the same way. It must recognize, for example, that an “address” field is likely the same as a “street address” field.
Gerken says that in order to work for businesses, Mashups also had to be designed to allow administrators to monitor what happens to the company’s data. For example, a popular mashup could strain a company’s database system if it constantly requested data from the same place. The system comes with features that help administrators notice this sort of traffic and respond to it by, for instance, noticing what information is popular and storing it in lighter, more easily accessible ways, Gerken says. Another potential problem with Mashups, he says, is that confidential company data often gets mixed with insecure, publicly available data. The mashup is a derivative product, he says, that might have different access requirements than its sources do. Gerken says that IBM has been looking at this problem through a research project called Damia and will include features from the research in Mashups. But, he says, the features aren’t finalized enough right now for him to elaborate on how, exactly, the product secures data at this point.
Niall Kennedy, a widget consultant in San Francisco, says, “There’s always been a demand for mashups, but the problem has been with the tools available for the interface.” He notes that before Google released tools for developers to use to make mashups with its map data, many developers reverse-engineered those tools in order to build their own applications. Lotus Mashups, Kennedy says, follows “a general trend of things getting adopted in the consumer space before getting a try in the enterprise.”
IBM plans to release Lotus Mashups in the middle of this year.