Member Junction
    Preparing search index...

    Interface SearchResultItem

    A single search result with provenance and scoring information.

    interface SearchResultItem {
        EntityDisplayName?: string;
        EntityIcon?: string;
        EntityName: string;
        FusionLane?: string;
        ID: string;
        MatchedAt: Date;
        PromotedFromContentItemID?: string;
        ProviderIcon?: string;
        ProviderId?: string;
        ProviderLabel?: string;
        RawMetadata?: string;
        RecordID: string;
        RecordName?: string;
        ResultType: SearchResultType;
        Score: number;
        ScoreBreakdown: SearchScoreBreakdown;
        Snippet: string;
        SourceType: string;
        Tags: string[];
        Title: string;
    }
    Index

    Properties

    EntityDisplayName? EntityIcon? EntityName FusionLane? ID MatchedAt PromotedFromContentItemID? ProviderIcon? ProviderId? ProviderLabel? RawMetadata? RecordID RecordName? ResultType Score ScoreBreakdown Snippet SourceType Tags Title

    Properties

    EntityDisplayName?: string

    Human-readable display name for the entity (e.g., "People" instead of "MJ_BizApps_Common: People")

    EntityIcon?: string

    Font Awesome icon class for the entity

    EntityName: string

    The entity this result came from

    FusionLane?: string

    Ranked-list key WITHIN a provider's results. Fusion ranks each distinct key as its own RRF list, so results whose scores are not comparable are merged by rank, never by score. The vector provider sets it to the embedding model: cosine scales differ by model (gte-small scores weak matches ~0.9, Gemini strong ones ~0.85), so one model's numbers say nothing about another's.

    Contract: a provider tags ALL of its results or NONE. Fully untagged = one ranked list. If a list is only partly tagged, the untagged remainder becomes its own lane rather than being merged into a tagged lane by raw score, which would compare scales again.

    ID: string

    Primary key of the source record (same as RecordID)

    MatchedAt: Date

    When this result was matched

    PromotedFromContentItemID?: string

    Set by the engine when it promoted a MJ: Content Items hit to the record the content was derived from: the content item's RecordID. EntityName / RecordID then name the origin record, which the provider in ProviderId never read — so none of that provider's guarantees carry over: the permission pass verifies the origin row (PK IN (...)), and the origin entity's lane ExtraFilter is re-checked. ProviderId, ProviderLabel and ProviderIcon still attribute the hit to the provider that found it.

    ProviderIcon?: string

    Font Awesome icon class from the SearchProvider metadata (e.g., "fa-solid fa-brain")

    ProviderId?: string

    ID of the SearchProvider metadata record that produced this result

    ProviderLabel?: string

    Display label from the SearchProvider metadata (e.g., "Database", "Semantic Search")

    RawMetadata?: string

    Raw vector metadata as JSON string (contains all entity fields stored in vector DB)

    RecordID: string

    The source record ID

    RecordName?: string

    Resolved display name for the record

    ResultType: SearchResultType

    Discriminator for UI rendering: entity-record, storage-file, or content-item

    Score: number

    Ranking score, 0-1: the RRF score divided by its maximum (1.0 = every lane that returned results ranked it first), or a reranker's relevance when one ran. Results are ordered by it. It is rank-based, not a calibrated relevance or confidence; each lane's raw score is in ScoreBreakdown.

    ScoreBreakdown: SearchScoreBreakdown

    Per-source score breakdown

    Snippet: string

    Text snippet showing the relevant content

    SourceType: string

    How the content was sourced: 'vector', 'fulltext', 'entity', or 'fused'

    Tags: string[]

    Tags associated with this result

    Title: string

    Display title for the result

    Settings

    Member Visibility

    On This Page

    Properties