Beginner’s Crash Course to Elastic Stack - Part 3: Full text queries

This workshop is Part 3 of the Beginner’s Crash Course to Elastic Stack.

Beginner’s Crash Course is a series of workshops for all developers with little to no experience with Elasticsearch and Kibana or those who could use a refresher.

In addition to full text queries, we will cover neat tips and tricks to build a combination of queries to answer more complex questions. We will wrap up by learning how to fine-tune relevancy using more complex queries as well!

Time Stamp:
- Part 2 review 1:46
- Match query 7:00
- Aggregations request 16:54
- Searching for a phrase with the match query 18:11
- Searching for a phrase with the match_phrase query 21:24
- Multi_match query 23:25
- Per-field boosting 25:27
- Bool query 31:41
- Must clause 36:50
- Must_not clause 38:58
- Should clause 40:13
- Filter clause 42:27

Resources
Table of Contents for Beginner’s Crash Course to Elastic Stack:

The link above includes the table of contents for all repos shared during the course.

These repos contain all the resources shared during the workshops including the presentation slides, the recording, as well as Elasticsearch requests.

Be sure to check out Part 3 repo as you follow along!

[rusyst]

Great stuff, guys, please make more of these. Lisa is such a great teacher!

[Alberal]

I really love the way Lisa talks about ES and how she explains all the concepts. Everything is crystal clear and so far this course has been super useful

[shine-odtsolmon]

Lisa is a great teacher. Thank you very much lisa

[t]

Eagerly waiting for part 4. Brilliant presentation.

[neelshah]

Thank you Lisa for workshop, especially revising previous topics about precision, recall, scoring/ranking did pave a road for an easier understanding of ES Full Text Queries! Cheers 👊

[thinker]

Thanks for these great videos, they're extremely helpful!

I did want to highlight one thing I found slightly confusing—at 41:40 you mention that it makes sense that:

query: bool: must: match_phrase: headline: Michelle Obama, and should: match_phrase: category: BLACK VOICES

returns the same # of results as the previous:

query: bool: must: match_phrase: headline: Michelle Obama, and must_not: match: category: WEDDINGS

but reflecting on it, it seems to me to be a random coincidence that they return the same number.

Logically, the query with the "must_not" should return fewer (lte the number of) results (ie have lower recall) than the previous query which returns all headlines with "Michelle Obama" and just uses "should" to change the score. If the number of results were the same, that would mean that there happened not to be any WEDDINGS articles with "Michelle Obama" in the headline, which would be a coincidence.

When I do this workshop using the data downloaded on 11/29/2022 (which has grown since your workshop), I get 220 for the "should match phrase category black voices" query and only 216 for the "must not match category weddings" query, suggesting that there are 4 WEDDINGS articles w/ "Michelle Obama" in the headline. I was able to verify this with a query: bool: must: [ { match_phrase headline Michelle Obama }, { match category Weddings } ]—it got 4 hits.

I think the point you're trying to make (that adding the "should" just changes scores and therefore, the order of the results, but not the number) would be clearer if you changed the should match BLACK VOICES query to also include the must_not: [ { match category WEDDINGS } ] clause.

[romankokhatskyi]

Thanks for the tutorials, really helpful!

[muditsaxena]

You can put filter query inside must and still get the same result right?

[hamadkhan]

Please make a video or two on logstash too.. want to know how it fits in :)

[alimirzaei]

Hi, thanks for nice presentation.
What is difference between must and filter?

[arfathali]

Amazing teaching and the format is beautiful easy to follow. <thumbsup>

[chauphuoc]

Are you going to make new video on this playlist ? Thks for helpful videos

[lingjiezhu]

Hi, thanks for the great videos. May I ask what are the difference between searching using the dev tools and from the console directly?