row upon row of clay pots in India

What Do We Mean By ‘Search’?

5 minute read
Martin White avatar
The word search has become a generic descriptor for "interactive information retrieval." In reality, there are a number of different categories of search.

The word "search" has become a generic descriptor for what the academic community describes as "interactive information retrieval." In reality, search encompasses a number of different categories. 

Understanding the features of each is critical because the techniques and technologies that improve the relevancy performance for one might not work for another. A search vendor recently used the example of how a query for NY in Google brings up a wealth of information on New York to suggest its AI-enabled software could do the same for enterprise search — which is doubtful, to say the least. For the purposes of this column I have briefly summarized nine categories of search.

1. WWW Search

Google and Bing dominate this area, but there are many others. These services all have to cope with massive volumes of content and user queries and yet deliver in less than 500 milliseconds. The query volume is a major factor in generating personalized results. To take the New York example, Google's machine data shows a high percentage of user queries for NY resulted in clicks on content about New York. 

It should also be noted that people and organizations construct websites to be found, so most of the content is curated to some extent, and that improves findability.

Related Article: The Essential Glue for Your Digital Workplace

2. Website Search

When it comes to content curation, websites do it well because good content reflects on the brand strength of the organization, no matter its sector or size. Sadly, few organizations pay any attention to site search, relying instead on the perceived quality (having been told by the design team) of the site's information architecture.

3. Intranet Search

The temptation with intranet search is to rely on the search application embedded in the CMS. Even when the vendor highlights the use of (for example) dtSearch or Solr, the implementation is rarely fit for purpose as no one understands the process of search, just the code base. An overreliance on the CMS really becomes noticeable when it becomes necessary to index and search other applications, such as an HR database. Searching across multiple languages is often a challenge.

4. Professional Search

Professional searchers have a need for astonishing levels of recall in order to challenge a patent application, win a court case or undertake a critical review of a medical procedure. These searchers use complex Boolean query strings to cope with synonyms and other aspects of the technical language of the discipline they are working in. Nowhere is good search more important to the objectives of the searcher and the organization. In general, they are dealing with highly curated content and often use specialized search applications such as LexisNexis.

Related Article: How Sticky Is Your Enterprise Search?

5. Ecommerce Search

Content curation is once again very high here. At the heart of ecommerce search is the concept of a stock-keeping unit (SKU) which defines a combination of price, functionality, manufacturer and much else. This data is held in a repository database and/or an enterprise resource planning application and needs to be surfaced in a way that a user can browse through sets of filters to create a short list of product options for purchase. This is where techniques such as learning-to-rank and a range of AI/ML algorithms can be very effective.

Learning Opportunities

6. Search-Based Applications

Many organization are using a search as the basis of a business application, such as a customer call-center or a recruitment service. These are closely related to ecommerce search, as there are limits to the number of different ways the data is going to be queried. For example, no one is going to build a list of potential candidates for a vacant position based on their birthday. These applications are typically highly customized and make extensive use of open source search applications, often developed internally.

7. Enterprise Applications

Many enterprise applications embed a more general-purpose search application into the code base. Examples would be enterprise resource planning (e.g. SAP HANA), HR, customer relationship management and building information management.

Related Article: Digital Workplace Success Relies on Strong Search

8. Enterprise Search

Enterprise search is arguably a complete outlier. The content is not curated to improve its findability by others and search security trimming might well result in employees at adjacent desks seeing different results. In a recent post I highlighted the three distinct use cases of search within the enterprise, with the long tail giving rise to a substantial number of issues. 

9. Departmental Search

Within an enterprise, individual departments may have implemented task-specific applications. A common example I've seen is pharmacovigilance, where the need to work through high volumes of customer feedback and external literature is a core business requirement.

Supporting Search

In terms of search support, the cross-category commonality of technical, user experience and user satisfaction management is limited. Each category needs its own optimization strategy and development roadmap. Most organizations manage these applications independently. In my view, a senior manager needs to have an overview of all of these applications in order to ensure that individually and collectively the organization is able to use them to achieve its objectives.

About the author

Martin White

Martin White is Managing Director of Intranet Focus, Ltd. and is based in Horsham, UK. An information scientist by profession, he has been involved in information retrieval and search for nearly four decades as a consultant, author and columnist.