Inverted Index Multi-Match Feature
multi-match is a powerful query feature that allows users to search a single query string across multiple fields simultaneously. This is very useful in many real-world scenarios. For example, when a user on an e-commerce website types "durable backpack" into the search box, the system may need to search across multiple fields such as product_title, description, and category at the same time.
When executing a multi-match query, the system performs matching operations on each specified field in the background, then intelligently merges and sorts all results, ultimately returning a unified relevance-ranked list.
Feature Example
Data Preparation:
Test data table:
Table structure:
id(string) - Entity IDtitle(string) - Entity titletext(string) - Entity description textvec(vector(float,1536)) - 1536-dimensional vector
Indexes already built:
| Index Name | Index Type | Target Field | Analyzer | Special Configuration |
| inverted_multi_match_idx_id | INVERTED | id | unicode | - |
| inverted_multi_match_idx_title | INVERTED | title | unicode | - |
| inverted_multi_match_idx_text | INVERTED | text | unicode | - |
| idx_dbpedia_vec_1536 | VECTOR | vec | - | ef.construction=128, m=64 |
Feature Examples:
Single-Column Match
Search in the title field, requiring 'Paris Wisconsin Foster' to match over 67%

Multi-Field Combined Search
Three-Field Combined Search (ID + Title + Text)
Search across id, title, text three fields, requiring at least 3 query terms to match
Result: 2 relevant records returned, with semantically accurate content matching.
