Semantic Search

Search content, users, and spaces with natural language queries

Sublay provides three semantic search hooks — one for content (entities, comments, messages), one for users, and one for spaces. All three return results ranked by semantic similarity to the query.

useSearchContent

Search across entities, comments, and chat messages. You can filter by sourceTypes to target only specific content types, or scope results to a specific space or conversation.

import { useSearchContent } from "@sublay/react-js";

function ContentSearch() {
  const { results, loading, search, reset } = useSearchContent();

  const handleSearch = async () => {
    await search({
      query: "posts about TypeScript best practices",
      sourceTypes: ["entity"],
      limit: 10,
    });
  };

  return (
    <div>
      <button onClick={handleSearch}>Search</button>
      {results.map((r) => (
        <div key={r.record.id}>
          [{r.sourceType}] {r.record.id} — similarity: {r.similarity.toFixed(2)}
        </div>
      ))}
    </div>
  );
}

Parameters (search function)

query · string · required

The natural language search query.

sourceTypes · ("entity" | "comment" | "message")[]

Limit results to specific content types. Defaults to all types.

spaceId · string

Scope results to a specific space.

conversationId · string

Scope results to a specific conversation (for messages).

limit · number

Maximum number of results to return.

Returns

results · ContentSearchResult[]

Ranked results.

Show ContentSearchResult properties

sourceType · "entity" | "comment" | "message"

The type of the matched record.

similarity · number

Cosine similarity score (0–1). Higher is more similar.

record · Entity | Comment | ChatMessage

The full matched record.

loading · boolean

true while a search is in progress.

error · string | null

Error message if the search failed.

search · (props) => Promise<void>

Executes the search and updates results.

reset · () => void

Clears results and error state.

---

useSearchUsers

Search for users by natural language query — matches against name, username, bio, and other profile text.

import { useSearchUsers } from "@sublay/react-js";

function UserSearch() {
  const { results, search } = useSearchUsers();

  return (
    <button onClick={() => search({ query: "designers in Berlin", limit: 5 })}>
      Search Users
    </button>
  );
}

Parameters (search function)

query · string · required

The natural language search query.

limit · number

Maximum number of results to return.

Returns

results · UserSearchResult[]

Ranked user results.

Show UserSearchResult properties

similarity · number

Cosine similarity score (0–1).

record · User

The matched user profile. See User data model.

loading · boolean

error · string | null

search · (props) => Promise<void>

reset · () => void

---

useSearchSpaces

Search for spaces by name, description, or topic.

import { useSearchSpaces } from "@sublay/react-js";

function SpaceSearch() {
  const { results, search } = useSearchSpaces();

  return (
    <button onClick={() => search({ query: "open source communities", limit: 5 })}>
      Find Spaces
    </button>
  );
}

Parameters (search function)

query · string · required

The natural language search query.

limit · number

Maximum number of results to return.

Returns

results · SpaceSearchResult[]

Ranked space results.

Show SpaceSearchResult properties

similarity · number

Cosine similarity score (0–1).

record · Space

The matched space. See Space data model.

loading · boolean

error · string | null

search · (props) => Promise<void>

reset · () => void