Improve search in Knowledge Base KB

[kevwalsh]

Background

This is a followup to the MVP that was developed in #5705

User Story or Problem Statement

As a KB user, i want to be able to find solutions to my problems, or answers to my questions, about the CMS and the product i edit.

Currently, there are a number of places where keywords are not surfacing the best result.

See this slack thread for some further reporting and possible solutions, and the first comment below.

ACs

• Work with user support team to find a number of unsuccessful searches
  • Document as many as possible (one bad search is not enough data to make a decision)
• Identify ways to improve KB search
  • Document ideas as discrete solutions
  • Recommend a subset of these for next steps.
  • Any recommendations to overhaul search architecture CMS should go into Improve content search tools #1196.

CMS Team

Please leave only the team that will do this work selected. If you're not sure, it's fine to leave both selected.

• Platform CMS Team
• Sitewide CMS Team
  • ⭐️ Content ops
  • ⭐️ CMS experience
  • ⭐️ Offices
  • ⭐️ Product support
  • ⭐️ User support

[stefaniefgray]

Current search is still troublesome:

When searching for "System Health Service," the articles about adding and editing a VAMC System Health Service should appear first.

See example: https://prod.cms.va.gov/help/search?keyword=system+health+service

Instead, the article "How to add a VAMC Facility Health Service" is listed before those -- and it is the 7th result! This is not even the correct type of health service.

The article "How to add a VAMC System Health Service" is the 14th result, and "How to edit a VAMC System Health Service" is the 16th result.

It is worth noting that the results do not improve in terms of relevance if you search for "System Health Service" within quotation marks, as they do when using search engines such as Google.

Ultimately, the current search functionality may make things very difficult for our Editors who expect to see the most relevant KB articles listed first.

Solution:

I believe that this issue can be rectified by having search prioritize the string keyword1+keyword2+keyword3 instead of searching for each word separately (which appears to be what the current search functionality is doing).

By prioritizing the phrase system+health+service instead of searching separately for instances of "system," "health," and "service" appearing on the same page, the Knowledge Base search functionality should be able to pull up the most relevant results first.

Lastly, KB search should also prioritize whether "keyword1+keyword2+keyword3" appears within the article title.

[msbtterswrth]

List of possible use cases:

1. As a CMS user, I want to find articles about log in help, but the page I want is not at the top of the list. (it is currently 5th)
2. As a CMS user, I want to search for “system health service” and find the relevant page about how to create a system health service. (https://prod.cms.va.gov/help/search?keyword=system+health+service drives me a bit crazy because “How to add a VAMC System Health Service” is the 14th result, yet the one about editing a VAMC Facility Health Service, the wrong content type, is 7th!)
3. As a CMS user, I want search results to be sorted by relevance with title (partial string) weighted highest
4. As an editor working on the 5 required VAMC trainings, the helpdesk has just informed me that I’ve completed every session except Session 3.Except… when I search for it in the KB Session 5 and Session 2 are listed, amongst a bunch of other pages, but I can’t find Session 3 at all! https://prod.cms.va.gov/help/search?keyword=session+3
5. As an editor searching for the term “pdf” I expect to see info on how to add a PDF to my site. I expect to find https://prod.cms.va.gov/help/cms-basics/file-link But this page is not actually listed at all in the search results: https://prod.cms.va.gov/help/search?keyword=pdf
6. As an editor I want to see how to check for broken links, but "broken link" returns how to insert a link to an external site. The 3rd option is the relevant article however the title is "Link checker reports"

[msbtterswrth]

Several improvements we can make are:

1. add another sort (title ASC) to the results view. This will improve readability for serialized articles/make it easier to find results.
2. add tagging by category (curated taxonomy)
3. add tagging by keyword (open taxonomy)
4. exposing filters for category, product in the results view
5. adding clickable keyword tags to search result and full page view
6. potentially adding the rate module so users can rate what was helpful and adding that as a weighting criterion
7. switching to SOLR based search can further improve the search accuracy, string matching and weighting options that we have available

[kevwalsh]

I've moved some of this to #8136 so we can close this issue!