Join us on Tuesday, December 19th, at Improving for a special Dallas Elastic User Group holiday meetup. Eric Robbins (Sr. Customer Architect at Elastic) will present 'Zero to Hero with Semantic Search,' followed by networking and festive holiday treats and refreshments.

Please RSVP if you plan on attending so we can provide the venue with an accurate headcount.

Interested in giving a lightning talk? 

Have you ever considered presenting but think 30 or 45 minutes is too long to start with? This is an excellent opportunity to try it out and build your speaking profile. Please send us an email at olivia@elastic.co.

Date and Time:

Tuesday, December 19th, from 6:00-8:00 pm CT

Location:

Improving Dallas - 5445 Legacy Dr #100, Plano, TX 75024

Agenda:

• 6:00 pm: Doors open; say hi, grab a seat, and eat some food.
• 6:30 pm: Zero to Hero with Semantic Search with Eric Robbins (Sr. Customer Architect at Elastic) followed by Q&A
• 7:30 - 8:00 pm: Networking & refreshments

Talk Abstract:

In this session, you will see how to go from zero to hero with semantic search using Elasticsearch and ELSER.

We will use common, open tools--from Vim and curl to Logstash and Elasticsearch--to ingest several years of Dallas Police Department incident reports from the city's open data API (Socrata) for analysis in Kibana.

While ingesting the records, we will pass the full text incident description from the reports through the Elastic Learned Sparse EncodeR (ELSER) NLP model to enrich each record with machine-learned tokens from the content as encoded by the language model. Then, we will query Elasticsearch using a text expansion search for records based on the meaning of our search phrase, not its keywords.

You will see how a search for "bad day at work" results in records containing "susp called comps work 60 times in 1 day" or "employee left one door unlocked after leaving", both accurately describing a bad day at work, but also content where a lexical search based on keywords alone would fall short in returning relevant incident reports.

As a bonus, you will see a Python Flask API used as a microservice in front of Elasticsearch, fully instrumented with OpenTelemetry and Elastic APM.

Attendees will have access to a private GitHub Gist with all the artifacts necessary to recreate what is shown on their own.