BI, enterprise search start to merge

In recent discussions with clients, I have seen some interesting developments in the business intelligence (BI) and enterprise search arena. These technologies are converging, a development that could have far-reaching implications into the business community and operational landscapes.

Business intelligence - as a concept - deals with structured data extremely well. Analysis engines have been optimised to deliver results based on structured information from databases and similar repositories.

Even the consolidation of data from multiple systems is performed according to a structured algorithm. Structure, and how to interpret the relationships within that structure, is a fundamental tenet of business intelligence.

However, business intelligence is generally available only to the elite users inside organisations. Not every worker can benefit from the potential value of BI. Cost is often seen as a barrier to wider adoption.

Secondly, business intelligence systems can seem overwhelming complex for the average user to interpret. The challenge is not the technology, but the complexity in relationships between the various data elements and components in the enterprise.

While the tide is turning, and the tools to expose business intelligence content to users are becoming easier to use, the principle remains much the same, where most people are unable to really, tangibly, benefit from the technology.

While enterprise search is still relatively immature, it has been able to realise broad utilisation due to its simplicity of operation.

Grant Hodgkinson is MD of Mint.

Search technology is still a relatively new concept. The impact that companies like Google have made on our lives is also still new, although it might not seem that way.

Bringing search technology within the enterprise is an even newer concept. While enterprise search works well with unstructured content, the multiplicity of repositories of content, the intricacies of security and other issues have created a high barrier to entry.

But it is merely that: a barrier. Given the right approach, the results can be astounding, and have a positive impact on all users.

Enterprise search is unlike business intelligence, in that it is a tool that is pervasive to a far greater percentage of the user population. People who would have huge difficulty manipulating a data cube can type in a keyword to search for content.

While enterprise search is still relatively immature, it has been able to realise broad utilisation due to its simplicity of operation.

Historically, business intelligence and enterprise search technologies have operated in different landscapes.

Business intelligence is posited as the silver bullet to analyse vast tracts of structured data. Enterprise search has been targeted at helping users find the needle in their unstructured information haystack.

Ultimately, both technologies are there to help users find information, and to help people advance innovation. In many instances, organisations recognise the need for both technologies to play a part in what has become a complex strategic and operational landscape.

Enterprise search is stretching tentacles into the structured data space. Most search platforms can index structured repositories of content. Similarly, business intelligence vendors are considering interesting ways of implementing keyword-search-like simplicity into their applications.

This intersection of functionality is only set to grow.

Even today, this has the potential to impact the platforms being implemented by organisations. While the merger of the two technologies is possibly a way off, there are some considerations even today:

1. Neither technology is a silver bullet to finding everything. Structured data and unstructured information alike will need to be formatted and exposed properly for them to be of relevance in either solution.

2. The principle of iterative releases of functionality is important. It does not make sense to implement either technology in a big-bang approach. Work with the data and content islands that represent the most value-add.

3. If exposing structured data to enterprise search, the structures or relationships embodied in that data need to be configured into the search system in order for results to be made relevant. The more complex cube-type manipulations that users perform in the business intelligence arena are also required in the search space, albeit slightly differently focused.

4. When using unstructured content in any search or lookup function, the search results are best matched to expectations when the content has been exposed in a manner that allows it to be contextualised. This often involves cataloguing information effectively. In short, the need for structure never disappears.

5. Consider the user base carefully. Sometimes more complex search and interrogation methods are required and appropriate, and this will preclude some workers. Sometimes the value of being able to find information within structured data is apparent to all users.

* Grant Hodgkinson is MD of Mint.