DOCUMENT Magazine - June 2008 - (Page 21) they can discover correlations they might not have considered, such as connections among symptoms, causes and cures. The search engine presents a cloud of content clusters to which a document belongs. When a search for a disease reveals clusters having to do with symptoms or possible causes, the doctor might learn something new. Some other ways to find such connections among enterprise data include: >> BI interfaces will encourage discovery of additional data dimensions. Some vendors let workers query a data model — that is, conduct a guided search — to find product categories or attributes, like stock keeping units (SKUs) and SKU groupings, color, size, weight and others. With tools like these, an information worker can enter a few keywords to find dimensions in the database and, using a graphical interface, drill into the information that is wanted from a list of possibilities. The real magic is that workers can actually get access to information without a fully formed question in mind, effectively resolving one of the oldest problems in business intelligence: having to know exactly what question to ask to get a meaningful answer. >> Finding mentions of phrases, names, people and places in mountains of text. An emerging technological capability called “entity extraction” can pick out from unstructured text names of people, addresses, brand names or just about any recognizable text or numeric pattern. The linguistic algorithms used to detect these patterns become useful when combined with the analytical and reporting capabilities found in BI platforms. For example, a risk manager Forrester spoke with at a large pharmaceuticals company now monitors corporate communications, blog sites and consumer discussion forums for any mention of adverse effects associated with its pharmaceutical brands. >> Understanding the overall tone of what people are writing. A technique called “sentiment analysis” uses similar techniques to evaluate the overall tone of what a writer is saying. Combined with traditional BI, sentiment analysis has the potential to drive new levels of insight in Voice of the Customer applications. For example, one upstart text analytics vendor that is helping a large hotel chain in North America correlate negative sentiments expressed in customer comments with specific cleanliness and food quality categories in its hotel chains in various regions of the country. “Previously, we hired teams of readers to painstakingly read comments in hundreds of feedback forms, which only includes a fraction of what our customers say about us,” explained the director in charge of the hotel chain’s Voice of the Customer program. What’s Next? In two short years, we’ve seen a flurry of related mergers and acquisition activities. We expect more to come as vendors seek to combine traditional BI capabilities with search and linguistics technology for everything from marketing applications to information risk management solutions. As search and BI get ever closer, the lines could eventually blur to the point of simply going away. An IT organization would then deploy a single application as an information access platform that would offer the robust analytics, reporting and visualization tools of BI combined with the deep linguistics processing and pattern matching of search. Boris Evelson is a principal analyst at Forrester Research. He is a leading expert in Business Intelligence. For more information or to view additional research written by Mr. Evelson, please visit www.forrester.com/ rb/analyst/boris_evelson. ■ www.DOCUMENTmedia.com june.08 document 21 http://www.forrester.com/rb/analyst/boris_evelson http://www.forrester.com/rb/analyst/boris_evelson http://www.DOCUMENTmedia.com
For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.