Overview

What Is Trusted Answer Search?

Oracle Trusted Answer Search is a purpose-built language-to-target mapping system for enterprise applications. It maps natural-language user questions to curated, application-defined search targets, such as reports, URLs, or SQL-driven views.

Trusted Answer Search is designed for scenarios where accuracy, speed, security, and repeatability are required. Instead of generating free-form responses, the platform returns deterministic results that the application can use to render the correct report or direct the user to the right page.

Trusted Answer Search combines Oracle AI Vector Search, lexical search, and re-ranking techniques to improve relevance. It also incorporates feedback-assisted multi-vector search so administrators can incorporate human feedback and continuously improve quality over time.

The platform includes operational controls for enterprise deployment, including change management, versioning, query regression analysis, and staged testing.

The Trusted Answer Search platform is intended to be integrated into applications that want to expose a natural language interface to their users to access the application’s functionality and reports.

Trusted Search Versus Agentic AI

The Enterprise Requirement

Many application experiences require deterministic and auditable outcomes. Users expect the same question to produce the same result, and organizations must ensure confidential data is never exposed.

Limits of General Agentic Patterns for Bounded Tasks

General-purpose Large-Language Model (LLM) and Retrieval-Augmented Generation (RAG) approaches are effective for broad, open-ended questions. However, for bounded, application-specific questions, they can introduce the following issues:

Trusted Answer Search Approach

Trusted Answer Search addresses these issues by using constrained retrieval and controlled outputs:

The result is a practical, enterprise pattern for trusted, repeatable language interfaces.

Architecture and Concepts

Architecture

Trusted Answer Search includes five components:

Personas and Access


Trusted Answer Search Architecture

Core Concepts

Search Target

A Search Target is a curated asset, such as a URL to a report, chart, or SQL-generated view. It is defined by descriptions and sample queries, and is what the system retrieves in response to user questions. Trusted Answer Search does not store or index the reports or URLs in any way. It simply returns the target to the application that uses Trusted Answer Search, along with any parameters contained in the natural language query. The application is free to take action (such as sending the user to the URL) based on the target returned.

Consider the example of a brokerage application that must answer common questions about current positions, previous trades, tax statement location, etc. Trusted Answer Search returns to the application the URL for the page showing current account positions (among other ranked results). The application or website then can redirect the user to the top returned result.

Following is a list of examples spanning diverse application domains:

Target Action

A Target Action is metadata associated with a Search Target. It is usually one of the following:

Trusted Answer Search returns action metadata to the client; execution is handled by the client application.

Continuing the brokerage example, the “Trade history in Q4” target publishes a deterministic action that downstream apps use to open a page displaying trade history with quarter filters applied.

Examples illustrate how applications consume target actions:

SELECT shift_id,
       start_time,
       end_time,
       required_headcount,
       scheduled_headcount
  FROM workforce_shifts
 WHERE region_code = :REGION_CODE
   AND start_time BETWEEN SYSDATE AND SYSDATE + 7
   AND scheduled_headcount < required_headcount
 ORDER BY start_time;

Target Inputs

Target Inputs make URL and SQL actions dynamic. Placeholders in an action template are populated from values extracted from user queries.

Example URL template:

https://www.brokerage123.com/investments?username=:USERNAMES

In this example, if the user query includes a username value, :USERNAMES is replaced with the recognized value.

So, Trusted Answer Search would return:

The Target Input :USERNAMES is linked to a Target Value Set MY_USERNAMES that contains a list of usernames so that Trusted Answer Search only substitutes approved values or synonyms. More on target value sets below.

Target Value Sets

Target Value Sets define allowed values used in extraction of keywords from a user query and substituting those extracted words into target actions. Two Target Value Set types are supported:

Synonyms can improve matching, while canonical values are used for action substitution.

In the brokerage example, the MY_USERNAMES Target Value Set might define canonical values such as JOHN_SMITH and ADAM_JONES with synonyms like “john” or “adam”. Target Inputs reference this value set so that :USERNAMES is replaced with the canonical user identifier even when the user query uses a synonym.

Search Space

A Search Space is a versioned collection of search targets for a specific application domain. Teams can isolate changes, collaborate, compare versions, and promote approved updates.

In the retail banking example, the Search Space expert maintains a “Retail Banking” search space that groups targets such as “Current account positions,” “Tax reporting statements,” and “Total return in the current FY.” Experts can stage updates, review diffs, and promote a tested configuration to production once approvals are complete.

How It Works

Trusted Answer Search uses a deterministic, LLM-free hybrid retrieval pipeline.

1. Query Normalization

The system normalizes user input before retrieval:

This produces normalized tokens and normalized query text.

2. Dual Retrieval

Semantic Retrieval

Vector similarity retrieval using Oracle AI Vector Search embeds query and corpus text to identify intent-level similarity. Corpus text includes target descriptions, target sample queries, and expert upvoted/downvoted user queries in search targets.

Semantic retrieval provides strong recall when wording differs.

Lexical Retrieval

Keyword-based matching retrieves targets from an inverted index built from:

Lexical retrieval provides high precision when terminology aligns.

3. Re-ranking and Fusion

Ranked lexical and semantic result lists are fused into a single ordered list using a re-ranking algorithm. This balances exact token match with semantic intent match.

4. Target Input Extraction

For targets with dynamic actions, the system matches normalized query terms against Target Value Set’s values and synonyms, then substitutes canonical values into URL or SQL templates.

Example Walkthrough

User query:

Show my stock trades in Q1 2026

  1. Normalize the query.

    • Spelling correction: No typos detected.
    • Morphology handling: Reduce “trades” to “trade,” remove auxiliary verbs.
      Normalized text: “show my stock trade in Q1 2026”
    • Stop-word filtering: Drop “show”, “my”, “in” and filler words to keep the high-value tokens.
      Normalized keywords: stock, trade, Q1, 2026.

  2. Run dual retrieval.

    Lexical search:

    • Matches detected: “stock,” “trade,” “Q1,” “2026” across target descriptions and sample queries.
    • Target Value Set exact matches: Q1 (quarter), 2026 (year).
    • Ranked lexical results:
      • Rank 1: “Same-Day Trade Confirmations” (precise hits on “trade,” “stock,” and the year token “2026”)
      • Rank 2: “Quarterly Stock Trade History”
      • Rank 3: “Equity Trade History – Year Selector”

    Semantic search:

    • Ranked semantic results:
      • Rank 1: “Quarterly Stock Trade History” (strong semantic match)
      • Rank 2: “Equity Transactions Timeline”
      • Rank 3: “Trade Confirmations Archive”

  3. Re-rank and fuse the lexical and semantic lists.

    • Rank 1: “Quarterly Stock Trade History” — despite ranking second lexically, it has strong lexical coverage and the best semantic match, so re-ranking promotes it to the top.
    • Rank 2: “Same-Day Trade Confirmations” — maintains a high position thanks to lexical precision, but trails the top result semantically.
    • Rank 3: “Equity Trade History – Year Selector” — retains solid lexical alignment yet weaker semantic relevance compared to the top two.

  4. Extract target inputs.

    For the “Quarterly Stock Trade History” target, the following target inputs are extracted from the user query:

    • Target action URL template: https://www.brokerage123.com/trade-history?quarter=:QUARTER&year=:YEAR&type=stock.
    • :QUARTER resolves to Q1.
    • :YEAR resolves to 2026.

The resulting URL is https://www.brokerage123.com/trade-history?quarter=q1&year=2026&type=stock, which the application can open directly to display the client’s Q1 2026 trade history.

This hybrid approach ensures customers get fast, consistent, and trustworthy answers to business questions, without relying on generative models.

Feedback-Aware Relevance

Trusted Answer Search incorporates feedback from the Search Space Expert to improve retrieval quality:

Reliability Characteristics