Separate language search, exact comparison and meaning expansion.
- Morphology and languages
- Multiple value representations
- Synonyms as controlled expansion
Separate language search, exact comparison and meaning expansion.
The course targets Elasticsearch 9.4.3. Exercise examples use the public products dataset; other requests illustrate separate teaching indices and may require setup or privileges unavailable in the trainer. Responses are shortened fragments; scores may vary. Check deployment and license requirements for advanced features.
Morphology and languages
A Russian analyzer can reduce inflected forms but does not understand intent like a person. Mixed languages, device models and brands need a test vocabulary. Separate language fields can make query targeting explicit. Transliteration is not automatic: it requires a deliberate representation and evaluation of false positives.
Check your understandingMultiple value representations
name can have a main text representation and a raw keyword subfield. One searches words; the other sorts or compares whole values. A keyword normalizer retains a single token and can normalize case; it is not a full analyzer. Adding a multi-field does not populate existing documents automatically: they need reprocessing.
Check your understandingSynonyms as controlled expansion
Equivalent and directional rules differ: “notebook, laptop” is not the same as “ultrabook => laptop”. Multiword synonyms require a token graph; synonym_graph is intended for search analysis. Correctly configured search-time synonyms can be updated without rebuilding all documents. Evaluate precision as well as increased result counts.
One source name gets separate full-text and exact representations.
{
"mappings": {
"properties": {
"name": {
"type": "text",
"analyzer": "russian",
"fields": {
"raw": {
"type": "keyword"
}
}
}
}
}
}How to read this result
One source name gets separate full-text and exact representations.
Catalog reasoning example
Inspect actual Russian analyzer tokens for inflected forms. Decide whether Sony and SONY are equal in exact filters and configure a normalizer accordingly. Russian and English equivalents need an explicit language or synonym strategy; morphology does not translate them. A synonym may introduce irrelevant accessories, so add positive and negative judgments for each change. A higher total alone is not a quality improvement.
Check your understandingCheck your understanding
Answer before expanding each explanation.
1. Is stemming translation?
No, it reduces word forms rather than translating languages.
2. Why use name.raw?
For exact values, sorting and aggregation independently of text analysis.
3. Why is a normalizer not a tokenizer?
It must retain a keyword value as one token.
4. Where should synonym_graph be used?
In search analysis for multiword alternatives.
5. What follows adding a multi-field?
Reprocess existing documents to populate the representation.
Key takeaways
- No, it reduces word forms rather than translating languages.
- For exact values, sorting and aggregation independently of text analysis.
- It must retain a keyword value as one token.
- In search analysis for multiword alternatives.
- Reprocess existing documents to populate the representation.
Official reference
Ready to put this into practice?
Confirm reading to save your place. Reading and exercise progress are tracked separately in this browser using your guest cookie.