Boosting types - Amazon Q Business

Boosting types

Amazon Q Business offers two types of boosting: document attribute boosting and document attribute value boosting. This section outlines how these types of boosting work.

Note

To use the STRING and STRING_LIST type document attributes for boosting on the console and the API, they must be enabled for search using the DocumentAttributeConfiguration object of the UpdateIndex API operation. If you don't enable search on these attributes, you can't boost attributes of these data types on either the Amazon Q Business console or the API.

Boosting document attribute importance

You can boost document attributes to control the relative importance, or boosting level, of the field for end user queries. You can boost importance for all document attribute data types that are supported by Amazon Q—DATE, NUMBER, STRING, and STRING_LIST.

Note

Amazon Q Business automatically boosts the document title attribute to Low. You can change this value when you customize boosting.

If you choose to boost document attributes, you can also customize boosting in the following ways:

  • Boost duration – Specifies the time period over which a boost applies to a DATE type document attribute. For example, if you set boosting duration to 604,800 seconds (1 week) for the _created_at reserved attribute, documents created within the last week will be boosted.

    Generally, all documents inside the boosting duration will be given more importance over documents outside the boosting duration. Within the boosting duration, documents with more recent dates will be given more importance over documents with less recent dates.

    Outside the boosting duration, the documents with more recent dates will continue to be given more importance over documents with less recent dates. However, the overall effect of the date boosting will taper to zero as the dates move further away from the boosting duration.

    Note

    Boosting duration is based on the most recent date in all documents in the index.

  • Boost order – Determines whether a NUMBER type document attribute is boosted in prioritizing higher values or prioritizing lower values.

    For example, if your documents contain attributes for view count, you can choose to prioritize chat responses with higher view count values by boosting larger values over smaller values. Or, suppose your documents contain attributes that denote priority—for example, a task tracker that assigns priority 1 to the most important task. In that case, you can choose to boost documents using smaller values.

Boosting document attribute value

To customize boosting levels, you can boost document attribute values for only STRING type document attributes.

For example, suppose that you're applying an importance boost to a STRING attribute called department. The department attribute has values like HR and Legal. You can assign the values HR, VERY_HIGH and Legal, HIGH to customize the importance that Amazon Q gives to these attribute values when they match a chat request.