# Semantic search

> Search the crawled site by meaning. The query is embedded on this computer and compared with every embedded page, so a page can match without sharing a word with the query. Each result shows the parts of the page that match best.

Describes GLLmolfar 0.9.22, checked 2026-11-01. Canonical page: https://www.getlooploop.com/gllmolfar/docs/semantic-search

- **Where:** Meaning › Semantic search
- **Mode:** Pro
- **Built from:** The open crawl's page and section embeddings
- **Changes:** On each search

## On the screen

A search box across the top, a line on how the search works, then the results.

![The Semantic search screen of GLLmolfar: the query how to measure AI visibility and the getlooploop.com pages that match it, each with its score and the sections that matched](https://www.getlooploop.com/images/gllmolfar/docs/semantic-search.webp)

Numbered parts of the screenshot:

1. **Find pages by meaning.** The query: a question or a phrase, in any wording. **×** clears it.
2. **Search.** Runs the search. Enter does the same.
3. **How it works.** **The query is embedded on this computer and compared with every embedded page.**
4. **What was compared.** How many pages the query was compared with, and how many had their sections searched: a result shows the part of the page that matched.
5. **The page's score.** The similarity of the page as a whole to the query, and whether the page as a whole or **a section matches**.
6. **Matching sections.** The sections of that page closest to the query, each with its own score and the start of its text.

| Part | What it shows |
| --- | --- |
| Find pages by meaning | The query: a question or a phrase, in any wording. |
| Search | Runs the search. Enter does the same. |
| How it works | The query is embedded on this computer and compared with every embedded page. |
| Results | How many pages were compared, then the closest pages, best first, or a message saying why there are none. |

## A result

Up to 25 pages, closest first. Above them: **Compared with N pages.**, and either **Sections searched on N pages: a result shows the part of the page that matched.** or, for a crawl made before 0.7.2, **This crawl has no section vectors (crawled before 0.7.2): crawl again to search by section.**

| Element | What it shows |
| --- | --- |
| Number | The result's place in the list. |
| Title and address | The page's title, or its address when it has none, and the full address with the colour of its topic. |
| Score | How close the page is to the query, from 0 to 1, with three decimals and a bar. The bar is empty at 0.20 and full at 0.80. |
| a section matches / the page as a whole | Which part gave the score: one section of the page scored higher than the whole page, or the page as a whole did. |
| Sections | Up to three sections of the page, best first: each with its score, its heading (**Start of the page** for text before the first heading) and the start of its text. |

A page scores as its best part: the whole page, or its best section. A result opens the page's details. The toolbar's **Export** saves the results on screen: rank, query, address, title, topic, similarity, best section and its text.

> **Reading the score.** The score is the similarity between the query and the page in the embedding model on this computer. It ranks the pages for this query; it is not a percentage.

## States and messages

- **No embeddings** — the app says: "No embedded pages yet." The open crawl has no embedded pages. Turn on **Embeddings** in the toolbar and let the model work through the pages.
- **Locked** — the app says: "Not included in your level. Open the tab or menu item to see what includes it." Printed in place of the results. A tab with a lock is not open on this account; pressing it says what opens it.

## Related pages

- [Topics](https://www.getlooploop.com/gllmolfar/docs/topics.md): The Topics tab: the site's pages grouped by meaning, with each topic's size, key terms, cohesion, internal linking, cannibal pairs and central page.
- [Topic map](https://www.getlooploop.com/gllmolfar/docs/topic-map.md): The Topic map tab: every embedded page as a dot, placed by meaning and coloured by topic, status, indexability or depth, with a card for each page.
- [Reverse prompts](https://www.getlooploop.com/gllmolfar/docs/reverse-prompts.md): Site › Reverse prompts: GLLmolfar reads a page into blocks, your AI agent describes what each does, and every block gets a reusable prompt and an example.
- [Settings › Embeddings](https://www.getlooploop.com/gllmolfar/docs/settings-embeddings.md): Settings › Embeddings: whether new crawls build embeddings, how much of each page the model reads, and whether it runs on the CPU or the GPU.