Business Objects Reaches into Unstructured Data
With new BI search and text analytic entries, Business Objects takes on unstructured data
- By Stephen Swoyer
- October 2, 2007
Business intelligence (BI) powerhouse Business Objects SA has shown itself to be become something of a pro at rapidly assimilating acquisitions.
Consider Inxight, which the company purchased just a few months ago. Last week, the firm formally closed the book on that acquisition, announcing two new products—BusinessObjects Text Analysis and BusinessObjects Intelligent Search—that incorporate Inxight technologies.
"Today we deliver a compelling solution to enable our customers to … discover, manage, and analyze unstructured content inside and outside of their organization," said Ian Hershey, vice-president of technology development and strategy at Business Objects, in a statement.
BusinessObjects Intelligent Search lets users comb structured, semi-structured, and unstructured data sources—including BI repositories as well as Web pages, RSS feeds, and company content, among other types of content—for apposite information. It’s also able to intelligently organize search results, filtering them in terms of relevancy. Its counterpart—BusinessObjects Text Analysis—is just what it sounds like, officials say: a text analytic bloodhound that’s able to search more than 220 text formats in over 30 different languages.
According to Business Objects, users can tap either tool to access and analyze textual information similarly to how they work with structured data now. In other words, according to company officials, both Enterprise Search and Text Analysis are tightly integrated with BusinessObjects Enterprise XI (the company’s BI suite) and BusinessObjects Data Integrator (the company’s ETL, or data integration, offering).
Similarly, both tools can be invoked from typical Business Objects client front-ends—for example, Crystal reports, Web Intelligence, Crystal Xcelsius dashboards—with a click of a button.
Tackling the Growth of Unstructured Data
Philip Russom, senior manager at TDWI Research, has a positive take on both tools, which he says address a salient—and (up to now) mostly neglected—requirement in most large organizations: a need to better organize and analyze unstructured data. According to TDWI Research, while the overwhelming majority (77 percent) of data consumed by data warehouses (DW) or BI processes is structured, unstructured data accounts for about one-third of all enterprise data. It just isn’t being consumed by DW or BI tools.
As organizations continue to unify and un-silo once disparate data sources, the amount of unstructured data being cataloged and fed into enterprise data warehouses will continue to grow.
"At least half of the data managed by corporate servers is unstructured, yet only a quarter of the content in the average data warehouse is based on unstructured data," Russom explains. "A data warehouse may be a single version of the truth, but it’s not the whole truth without information drawn from unstructured data. Technologies for search and text analytics can complete the data that decisions and reports are based on, thereby making these more accurate and useful."
At the same time, BI search has a specific set of requirements that are quite distinct from the ubiquitous Web search model, Russom indicates. In this respect, he argues, BusinessObjects Enterprise Search isn’t just an omnivorous search engine bolted on top of its BI and data integration stacks.
"Adapting search technology to BI applications is not trivial. For search to be analytic—which is what BI users want from any technology—it must provide an interpretation of the information it finds, not just list a ton of documents," he points out. "BusinessObjects Intelligent Search uses clustering techniques to categorize documents and information by people, place, and things, thereby giving the user a structure similar to a table of contents, instead of an undifferentiated list."
Ditto for Text Analysis, which incorporates technology assets Business Objects acquired just five months ago from the former Inxight.
"Turning textual information into something a data warehouse can understand is hard, but that’s exactly what text analytics does. It parses—or reads—human language and other text, and creates a database record or similar structure, based on what it finds," he notes.
"BusinessObjects Text Analysis can create these structures and hand them directly to BusinessObjects Data Integrator, which in turn transforms and loads them into a data warehouse, thereby helping to rectify the paucity of textual information suffered by most data warehouses."