At the present moment ‘search’ is being used by the AI community (especially in the context of ChatGPT) as a generic product. The reality is that there are at least eight facets of search, each with their own particular requirements.
A very brief (and inevitably superficial) summary of each is set out below. A future post will examine enterprise search in more detail.
Massive volumes of everything dominated by Google and the need for $$. People want their content to be found so pay attention to quality, metadata and links
Corporate web site
Highly curated content but search is so often seen as an incidental add-on as surely the information architecture is perfect!
Because of the diverse range of content in an intranet a good search feature is essential. As with corporate web site search the content is highly curated.
Services for (primarily) academic users with highly curated content on special-purpose applications
Highly curated content and very good user tracking metrics where revenue improvement drives search development
Users with excellent retrieval and subject skills (for example in legal practices) working with highly curated content and often a requirement for 100% recall
Users primarily in clinical research using complex queries against highly curated content and a requirement for 100% recall
Complex array of content types, often in different languages and sorted in a range of applications where the content is not curated against any corporate standards.
The theme running through all these search applications, with the specific exclusion of enterprise search, is that the author of the content wants it to be found and care is taken by the author and the organisation to ensure that the content is current and well-structured. In enterprise search the author is only interested in their contributions being discoverable by colleagues.
Martin White Principal Analyst