Documentation menu

Semantic search

Describes GLLmolfar 0.9.22current release 0.9.23 · checked 1 Nov 2026

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.

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
real screenshotProject getlooploop.com. Other sites and brands on screen are replaced with neutral names.

  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.

PartWhat it shows
Find pages by meaningThe query: a question or a phrase, in any wording.
SearchRuns the search. Enter does the same.
How it worksThe query is embedded on this computer and compared with every embedded page.
ResultsHow 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.

ElementWhat it shows
NumberThe result's place in the list.
Title and addressThe page's title, or its address when it has none, and the full address with the colour of its topic.
ScoreHow 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 wholeWhich part gave the score: one section of the page scored higher than the whole page, or the page as a whole did.
SectionsUp 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.

States and messages

No embeddingsNo embedded pages yet.

The open crawl has no embedded pages. Turn on Embeddings in the toolbar and let the model work through the pages.

LockedNot 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

TopicsThe 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 mapThe 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 promptsSite › 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 › EmbeddingsSettings › Embeddings: whether new crawls build embeddings, how much of each page the model reads, and whether it runs on the CPU or the GPU.