Skip to main content

Elasticsearch for Product Searches

Make Product Search Better

Rating 

Our Elasticsearch training classes have a 4.74/5 rating based on 20 reviews

This 2-hour online course focuses on E-commerce or similar product search use-cases (e.g. searching for phones, books, people, blog posts). Because relevancy is important for product search: Elasticsearch shouldn’t return “phone accessories” on a query for “phone”. Other challenges might come up as well, for example dealing with relational data, providing autocomplete or did-you-mean functionality. This course provides solutions to these challenges. Radu Gheorghe, a seasoned Elasticsearch instructor, and consultant from Sematext, author of “Elasticsearch in Action”, and frequent conference speaker will be your trainer.

Why attend?

  • Small, interactive, instructor-led classes
  • Lots of hands-on exercises
  • Customized learning experience
  • More flexible – no need to travel
  • Certificate of Completion included

What’s Included

  • 8-hour online training
  • A digital copy of the training material
  • Docker Compose files, configs, scripts, etc.
  • Certificate of Completion

Next Class: Sept 30 – Oct 1st, 2020 See Upcoming Classes

$800.00 Register Now

Who should attend?

This Elasticsearch online course is designed for anyone who:

  • Has a basic understanding of Elasticsearch’s core concepts (documents, shards, basic queries)
  • Is looking to implement product search.

Upcoming Classes

Pick from our 8h online classes, structured to correspond to different roles and Elasticsearch knowledge levels: for beginners to experienced developers or ops who want to learn quickly. Delivery method: Live Online. Time: 09:00 AM – 01:00 PM ET (2 sessions).

DateClassPriceRegistration
Sept 30 - Oct 1st, 2020Elasticsearch for Product Searches$800 / personSee Course Outline Register Now

What attendees say

Sematext was an ideal training partner for Parse.ly. We had just recently adopted Elasticsearch on a new project, and they gave us two days of solid training that was tailored to our team’s needs. The material was built atop strong foundations and moved quickly into advanced areas around querying, Lucene internals, and cluster performance. It was clear that it was all informed by real-world experience operating these systems at scale.

Andrew Montalenti CTO/Founder – Parse.ly

Course Outline

Making sure you get the results you want
  • Using ngrams and shingles to tolerate typos
  • Using stemming and synonyms to expand the scope of a search
  • Balance precision and recall by using different variants of the same field
  • Don’t cast a net that’s too wide: minimum_should_match and cross_fields
  • Lab
    • Controlling the number of terms matched by a query
    • Comparing ngrams and shingles with fuzzy and phrase searches
    • Dealing with stopwords efficiently
    • Implementing analysis for URL matching and hashtag search
Relevancy tuning
  • How is relevancy score calculated and how can it be tuned
  • Searching across fields to detect user intent
  • Using teh Function Score Query to boost documents by recency, region and other business criteria
  • Lab
    • Changing the way score is calculated for a field
    • Boosting more exact matches
    • Boosting documents based on recency
Using aggregations to boost search experience
  • Using terms aggregations to show facets
  • Showing relevant classes of products with the significant terms aggregation
  • Field collapsing and the top hits aggregation
  • Lab
    • Finding a balance between precision and performance for terms aggregation
    • Checking trends with the significant terms aggregation
    • Implementing results grouping
Relational data in Elasticsearch
  • Arrays and objects: fastest way to deal with one-to-one relations
  • Nested documents: quick searches for low-velocity data
  • Parent-child relationships: fastest way to update
  • Using inner hits to show the joined documents
  • Dealing with many-to-many relationships via application-side joins
  • Lab
    • Searching across a relationship between two documents
    • Showing both sides of the relationship in the results
Geo-spatial search
  • Indexing geo-point and geo-shape types
  • Bounding box and distance queries
  • Distance and grid aggregations
  • Shape query
  • Lab
    • Finding products near you
    • Finding products within an area
    • Faceting products based on distance from a pin
Suggesters
  • Term suggesters: providing did-you-mean for each word
  • Phrase suggester: did-you-mean for a whole query
  • Quick autocomplete with the completion suggester
  • Filter autocomplete results with the context suggester
  • Lab
    • Implementing did-you-mean functionality
    • Implementing autocomplete functionality
    • Add weights and fuzzy matching to type-ahead search
Highlighting query results
  • Default highlighting: fast and flexible for short fields
  • Term offsets: quick highlighting for content fields
  • Term vectors: paying with storage to get the best of both worlds
  • Lab
    • Add highlighting for query results
    • Choosing highlight tags and encoding
    • Choosing different modes for teh Unified Highlighter
Search turned upside down: Percolator
  • Registering alerts as queries
  • Running the percolator query to check which alert matches
  • Lab
    • Using Percolator to register price alerts on products
    • Filtering alerts by user and other metadata
 

Main Topics

  • Balance precision and recall through analysis tweaks
  • Using DisMax and Function Score to make relevant documents rank higher
  • Suggesters for did-you-mean and autocomplete functionality
  • Highlighting query results

Elasticsearch Training

Radu Gheorghe

About the trainer

Radu Gheorghe

Your trainer is an active Elasticsearch consultant. Radu has worked with clients from 20+ different industries and is the author of Elasticsearch in Action. Here are some problems that Radu solved for Sematext clients recently:

  • Improved search relevancy using Learning to Rank
  • Optimized multiple petabyte-scale clusters. Some up to 400 nodes.
  • Designed Elasticsearch index and cluster architecture for dozens of clients
  • Optimized log ingestion pipelines to parse and enrich 100K+ events/second
  • Helped clients reduce production Elasticsearch and ingestion pipeline costs by as much as 10x

Stay up to date

Get tips, how-tos, and news about Elastic / ELK Stack, Observability, Solr, and Sematext Cloud news and updates.

Sematext Newsletter