Filters don't work with when "Hybrid" is used as the Algo #131
Description
When executing a hybrid search query in Chorus by selecting the "Hybrid" algorithm in the dropdown box and entering a query term and filtering afterwards, nothing happens in the frontend.
In the browser dev tools it becomes clear that the backend has an issue:
{
"took": 21,
"responses": [
{
"error": {
"root_cause": [
{
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries"
}
],
"type": "search_phase_execution_exception",
"reason": "all shards failed",
"phase": "query",
"grouped": true,
"failed_shards": [
{
"shard": 0,
"index": "ecommerce",
"node": "uFWvZB3JQl-ZlKAcyeiTlg",
"reason": {
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries"
}
}
],
"caused_by": {
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries",
"caused_by": {
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries"
}
}
},
"status": 400
},
{
"error": {
"root_cause": [
{
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries"
}
],
"type": "search_phase_execution_exception",
"reason": "all shards failed",
"phase": "query",
"grouped": true,
"failed_shards": [
{
"shard": 0,
"index": "ecommerce",
"node": "uFWvZB3JQl-ZlKAcyeiTlg",
"reason": {
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries"
}
}
],
"caused_by": {
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries",
"caused_by": {
"type": "illegal_argument_exception",
"reason": "hybrid query must be a top level query and cannot be wrapped into other queries"
}
}
},
"status": 400
}
]
}
Query payload - how the query looks like that Reactivesearch is building:
GET ecommerce/_search
{
"query": {
"bool": {
"must": [
{
"bool": {
"must": [
{
"hybrid": {
"queries": [
{
"match": {
"title_text": {
"query": "claypottery"
}
}
},
{
"neural": {
"title_embedding": {
"query_text": "claypottery",
"k": 50
}
}
}
]
}
},
{
"bool": {
"should": [
{
"terms": {
"attrs.Brand.keyword": [
"Antallcky"
]
}
}
]
}
}
]
}
}
]
}
},
"size": 20,
"_source": {
"includes": [
"*"
],
"excludes": [
"title_embedding",
"reviews",
"description",
"bullets"
]
},
"search_pipeline": "hybrid-search-pipeline",
"ext": {
"ubi": {
"query_id": "00f48767-3938-4ee3-b836-eeb9203c4e18",
"user_query": "claypottery",
"client_id": "CLIENT-eeed-43de-959d-90e6040e84f9",
"object_id_field": "asin",
"application": "Chorus",
"query_attributes": {}
}
},
"aggs": {
"attrs.Brand.keyword": {
"terms": {
"field": "attrs.Brand.keyword",
"size": 20,
"order": {
"_count": "desc"
}
}
}
},
"from": 0
}
How the query should look like - with post filtering:
GET ecommerce/_search
{
"query": {
"hybrid": {
"queries": [
{
"match": {
"title_text": {
"query": "claypottery"
}
}
},
{
"neural": {
"title_embedding": {
"query_text": "claypottery",
"k": 50
}
}
}
]
}
},
"post_filter": {
"match": {
"attrs.Brand.keyword": "Antallcky"
}
},
"size": 20,
"_source": {
"includes": [
"*"
],
"excludes": [
"title_embedding",
"reviews",
"description",
"bullets"
]
},
"search_pipeline": "hybrid-search-pipeline",
"ext": {
"ubi": {
"query_id": "00f48767-3938-4ee3-b836-eeb9203c4e18",
"user_query": "claypottery",
"client_id": "CLIENT-eeed-43de-959d-90e6040e84f9",
"object_id_field": "asin",
"application": "Chorus",
"query_attributes": {}
}
},
"aggs": {
"attrs.Brand.keyword": {
"terms": {
"field": "attrs.Brand.keyword",
"size": 20,
"order": {
"_count": "desc"
}
}
}
},
"from": 0
}
Note: "Keyword" and "Neural" both work.