This topic describes the extended parameters of OpenSearch LLM-Based Conversational Search Edition.
Prompt parameters
attitude
Description: the tone of the answer. Default value: normal. Valid values:
normal
polite
patience
rule
Description: the level of detail of the answer. Default value: detailed. Valid values:
detailed: The answer is detailed and professional.
stepbystep: The answer is detailed and provides step-by-step instructions.
noanswer
Description: the response when the question cannot be answered. Default value: sorry. Valid values:
sorry: Sorry, I cannot answer your question based on known information.
uncertain: I do not know.
language
Description: the language of the answer. Default value: Chinese. Valid values:
Chinese
English
Thai
Korean
role
Description: the custom role that is used to answer the question. Example: AI assistant.
out_format
Description: the format in which to return the answer. Default value: text. Valid values:
text
table
list
markdown
Document retrieval parameters
filter
Description: the filter that is used to retrieve documents by field. By default, this parameter is left empty.
Syntax: field=value.
Examples:
1. Query data from the documents whose value of the category field is value1.
"filter" : "category=\"value1\""2. Query data from the documents whose value of the category field is value1 or value2.
"filter" : "category=\"value1\" OR category=\"value2\""3. Query data from the documents whose value of the category field is one of the specified values.
Separate multiple values with commas (,).
Example: category=value1,value2,value3,value4
"filter" : "category=\"value1,value2,value3,value4\"" // Retrieve the documents whose value of the category field is one of the specified values.sf
Description: the threshold for determining whether a retrieved document is relevant during vector-based retrieval. Default value: 1.3. A greater value specifies less relevance. Valid values: [0,+∞).
Syntax:
sf=value. Example: sf=1. You can change the value of the sf parameter based on your business requirements. A smaller value specifies greater relevance.
top_n
Description: the number of documents to be retrieved. Default value: 5. Valid values: (0,50].
Syntax:
top_n:value. Example: top_n:3. You can change the value of the top_n parameter based on your business requirements.formula
Description: the formula that is used to sort the retrieved documents.
Syntax:
Text relevance
text_relevance: calculates the text relevance between search queries and field values in documents.
field_match_ratio: returns the ratio of the number of terms in a field that match the search query to the total number of terms in the field.
query_match_ratio: returns the ratio of the number of terms that are hit in a field to the total number of terms in the search query.
fieldterm_proximity: returns the proximity of terms in a field.
field_length: returns the number of terms in a field.
query_term_count: returns the number of terms in the search query after analysis.
query_term_match_count: returns the number of terms in the search query that are hit in a field in documents.
field_term_match_count: returns the number of terms in a field that match the search query.
query_min_slide_window: returns the ratio of the number of terms in the search query that are hit in a field to the minimum window of these terms in the field.
Timeliness
timeliness: returns the timeliness score that indicates how new a document is. Unit: seconds.
timeliness_ms: returns the timeliness score that indicates how new a document is. Unit: milliseconds.
Functionality
tag_match: matches query clauses with documents based on tags and calculates the weights of matched tags to score the documents.
first_phase_score: returns the score that is calculated by using rough sort expressions.
kvpairs_value: returns the value of the specified field in a kvpairs clause in a query string.
normalize: normalizes scores in different value ranges to numeric values in the range of [0,1].
in or notin: checks whether field values are in or not in the specified list.
operator
Description: the operator that is used to define the logical relationship between tokens used to retrieve documents. Default value: AND. Valid values: AND and OR.
Manual intervention parameters
sf
Description: the threshold for triggering manual intervention. Default value: 0.3. Valid values: [0,+∞). A greater value specifies that a manual intervention entry is more likely to be matched.
Syntax:
sf=value. Example: sf=1. You can change the value of the sf parameter based on your business requirements.
Reference image parameters
sf
Description: the threshold for determining the vector similarity between the reference image and the specified content. Default value: 1. Valid values: [0,+∞). A greater value specifies less vector similarity.
Syntax:
sf=value. Example: sf=1. You can change the value of the sf parameter based on your business requirements.
Other parameters
return_hits
Description: specifies whether to return the search results. Default value: false, which specifies that only reference links are returned.
Syntax:
return_hits:value. Valid values: true and false. Example: return_hits:true. If you set the return_hits parameter to true, the corresponding search results are returned.csi_level
Description: specifies whether to moderate the results that are generated by large language models (LLMs) for restricted content such as sensitive, political, or harmful content. Valid values:
none: does not moderate the results.
loose: moderates the results and blocks the results if restricted content is detected. In this case, no results are returned.
strict: moderates the results and blocks the results if restricted or suspicious content is detected. In this case, no results are returned.
link
Description: specifies whether the reference source is included in the content generated by a model. Valid values:
true: The reference source is included in the content generated by a model.
false: The reference source is not included in the content generated by a model.
Sample response if you set this parameter to true:
You can resize the disk of an Elastic Compute Service (ECS) instance online or offline[^1^]. If you use the online resizing method, you can resize the disk without the need to restart the instance. If you use the offline resizing method, you must restart the instance[^1^]. To resize a disk, perform the following steps: Log on to the ECS console, find the disk that you want to resize, click Resize in the Actions column, and then select a resizing method based on your business requirements[^1^]. If you need to resize partitions and file systems, you can obtain relevant information by using the CLI or in the console[^2^]. After an ECS disk is resized, you cannot reduce the capacity. We recommend that you implement reasonable capacity planning[^3^].[^Number^] indicates the ordinal number of the retrieved document in the reference of the returned results. For example, [^1^] indicates the first document in the reference.