Effective search is a difficult problem to solve for most organizations. The challenge of tackling complex interdependencies between source systems and trying to satisfy stakeholders with different search needs can be daunting. Unfortunately, poor search isn’t just an inconvenience, it’s a liability. The inability to find critical information when it’s needed most diminishes trust in systems impedes employee learning and leads to less than optimal decisions.
The good news is, these problems can be solved.
Search-based applications enter the findability fray promising to shape enterprise search in a manner uniquely suited to tame the information explosion challenge. These applications have the inherent ability to aggregate information from multiple systems while applying contextually relevant domain knowledge, allowing them to answer questions not readily available from any single source system.
This new breed of application is made possible by the evolution of search technology beyond basic keyword search to include relevancy ranking, concept searching, text analytics, clustering, entity extraction and social interaction. While a traditional, well-implemented search engine is critical in helping employees make better use of corporate resources, in many cases benefits can be markedly improved by layering a domain specific search-based application on top of the core search infrastructure.
The Lone Search Box
One of the central challenges with traditional search is that the keywords a user picks to search on, the very question that is being asked, shapes the form of the answer. That’s fine if users know exactly what they are looking for, however, in many instances they may not have all of the information necessary to ask the right question. Ask too narrow of a question and the results resemble tunnel vision, ask too broad of a question and the results loose focus. In addition to keyword search, users need a mechanism that enables relevant discovery when the question isn’t quite so clear.
The trouble with the ever-present search box tucked away in the corner of a traditional web or desktop application is that it typically provides little or no context. Search that includes the context of who a user is and what they are doing at the time of search greatly enhances the relevancy of returned results. For example, say a user is currently viewing a web page describing the details of a car they are considering purchasing. The web page doesn’t explicitly mention warranty information so the user enters the term “warranty” in the “contextually unaware” search box on the page. Because the user wasn’t very specific in selecting their search terms and because the underlying search engine wasn’t aware of what car the user was viewing at the time of the search, chances are that the search results won’t be very targeted or relevant. However if the search infrastructure enriched the query with contextually relevant information about the year, make and model of the car being researched the search results are likely to be much more relevant.
A Dynamic, Adaptive Model
But why stop there? What if you never had to ask the question in the first place! What if the web page had all of the relevant information the user wanted right from the start? That’s a difficult promise to fulfill if the page has been built by hand or with a traditional content management solution. Building and maintaining links to all relevant related pages and documents can be a maintenance nightmare. However a search-based application has the intelligence to make both the end-user experience more productive and the job of maintaining the application much more tenable. That’s because by using search as the underlying platform for the application, links and relationships are discovered and maintained automatically. Search results can iteratively feed other searches behind the scenes and the whole system can adapt to the user’s changing environment in real-time. This approach is a fundamental move away from statically built applications into adaptive systems. Dynamic adaptability is critical to enhanced findability because the truth of the matter is that search is not a one size fits all solution. It should adapt to both the profile of the user and the context of the domain.
In the final analysis, a well designed, search-based application combines these critical factors into a single application, the ability to perform targeted keyword searches when you know exactly what you are looking for and dynamic exploratory search presenting serendipitous relevant information or paths to information when you aren’t. Automatically enrich these search options with contextual domain information and your users will begin to see search in a new light.
clustering, concept searching, contextual intelligence, dynamic adaptability, enterprise search, entity extraction, information connectivity, Information Optimization, relevance, relevancy ranking, social interaction, text analytics
Thanks for your article. I referenced it from my blog post: http://readrz.blogspot.com/2012/01/what-is-exploratory-search.html