Why Precision vs. Accuracy Matters with Your Knowledge Base

March, 9 2018

Subscribe and stay up to date

No spam, we promise! You will only 
receive essential emails.
Written By
Juliette Kopecky

The most important thing to understand about search engines is that there is a difference between being precise and being accurate, which is to say there's a difference between being correct and being helpful. If your knowledge base can't make that distinction, it will have limited usefulness to your organization.

And most knowledge base solutions can't really be helpful without artificial intelligence.

The gold standard for search engines today is Google, but the underlying mechanism of Google Search can't really be applied to most knowledge bases. The Google search algorithm was originally designed to rank the authority of scientific and academic publications by seeing how well and how often those papers were cited by other research. The more often a paper was cited, the more authority it was deemed to have. Under the original Google algorithm, when you search for a term, keyword, or phrase -- like "cell mitosis" or "cold fusion" -- Google will return items that are related to that phrase, ranked in order of their authority.

The Google web search simply replaces academic papers with websites, and citations with hyperlinks. Web pages that are most linked to a certain phrase show up first in Google Search results. This works because there are over a billion sites on the web today, all massively cross-linked, giving Google enough data to accurately provide relevant results for almost any search term.

Your internal knowledge base probably doesn't have a billion content items, and it certainly doesn't have decades of cross-linked citations that can power a Google-style search algorithm. Thus, simply having a search term present in a knowledge base article is usually enough to have it show up in an internal search -- whether that search result is actually helpful or not.

More simply, just because an entry in your knowledge base mentions the phrase "employee handbook" a great deal doesn't mean it is the employee handbook. In fact, the employee handbook probably only includes the phrase "employee handbook" on its title page, so it may be the least likely result for the search term "employee handbook."

It may be correct to return a document that includes "refer to employee handbook" 15 times when you search for "employee handbook", but it's not helpful. What's helpful is the Employee Handbook.

Artificial intelligence, however, can apply context to knowledge base items to improve their search-readiness. An A.I.-enhance knowledge base could automatically tag an Employee handbook entry with useful meta-data that ensures it shows up on relevant search, and could also prompt your documentation team to manually tag it with better terms. A.I. can also go the extra mile and note when multiple versions of the same document exists, and learn to surface the more relevant or up-to-date ones -- as well as prompt your staff to curate the content better (and maybe remove or archive the old versions).

In simplest terms, A.I. can act like an editor or librarian for your knowledge base, filling in the gaps that a "dumb' search algorithm can't -- especially when that algorithm doesn't have enough data to build a robust search index.

If you'd like more information on how artificial intelligence can create a better knowledge base, contact Talla today.

New call-to-action

View all posts

Recommended Reading

Subscribe and stay up to date

No spam, we promise! You will only 
receive essential emails.

Subscribe and stay up to date