Relevant search results (with elasticsearch)
summary
A demo search app indexes the Rolling Stone top 500 albums (500 records) and uses Java, Angular and Elasticsearch via the Kibana console to show concrete relevance techniques: bool queries with function_score linear decay on release_date to boost recent albums, match and multi_match with field boosts, analyzer pipeline (char filter → tokenizer → token filters) and mapping with keyword vs text multi-fields, TF‑IDF scoring visible through the explain API, edge ngrams for autocomplete, and a manual synonyms file (a synonym demo fails during a rebuild).
AI-written summary · about
chapters (22)
- 0:00Introduction to relevant search
- 2:45What are relevant search results
- 5:58Content and features in search
- 7:01Signals and ranking in Elasticsearch
- 8:05TF-IDF scoring explained
- 11:20Why use a search engine like Elasticsearch
- 14:12How inverted indexes work
- 15:57Text analysis in Elasticsearch
- 18:32Understanding mapping in Elasticsearch
- 20:41Basic queries in Elasticsearch
- 21:46Match queries and filters
- 23:11Boosting fields for better relevance
- 25:19Function score queries
- 27:09Boosting by release date
- 30:41Synonyms in Elasticsearch
- 34:00Edge n-grams for suggestions
- 35:45Q&A: Filter order and quotes
- 36:08Q&A: Handling confidential data
- 36:50Q&A: Masking search results
- 37:12Q&A: Automating synonyms
- 37:54Q&A: Elasticsearch vs Solr
- 38:37Closing remarks
related talks
01h 48mInformation Retrieval from the Ground Up
30mSearch on Steroids — A Story of Semantics
44mSearching through large graphs using Elasticsearch
15mHow Big Data Makes Content More Relevant in Search
14mSearch on Steroids
19mYou Know, for Search. Querying 24 Billion Records in 900ms
42mResearch in the Age of AI
34mEvolution of Slack Search