# Information retrieval built the machine that marketing talks about without reading its manual.

Canonical: https://www.searchresultoptimization.com/information-retrieval/
Author: Johannes Faupel

On this page

**What it is**
Information retrieval is the science of finding the document, and today the passage, that answers a need.

**The problem until now**
Its measures stayed inside the field, so nobody read their own pages the way a retrieval system reads them.

**What you can do from here**
You can see which passage of yours can be found, scored and taken, and what weakens it.

![Information retrieval: a page is split into passages, passages are scored for a query, the ranked list follows](https://www.searchresultoptimization.com/images/maps/passage-retrieval-information-retrieval.svg?v=1790368036)
*Information retrieval: How the parts depend on each other.*

## Try this first
<https://www.searchresultoptimization.com/information-retrieval/#try-this-first>

Three tests, about ten minutes. They show your pages as a retrieval system sees them.

Tests on this page you have run: 0 of 3. The count stays in this browser only.

Test: Information retrieval

- Take one heading of your page that is phrased as a question.
- Read only the first sentence under it.
- Think: does that sentence answer on its own?
Whether a passage can be taken out and still work.

Test

- Search your topic and note the exact words the top results use.
- Note what you got, before you open anything else.
- Think: which of those words are missing from your page?
The vocabulary a retrieval system counts.

Test

- Write down five facts of yours that should be findable.
- Search each one with `site:example.com` and your own domain.
- Think: how many were found?
Your own recall, out of five.

Search engines are engineering before they are marketing. The people who optimize search engines are the engineers who build them.

## Topics in this field
<https://www.searchresultoptimization.com/information-retrieval/#topics-in-this-field>

[Learning to rank](https://www.searchresultoptimization.com/information-retrieval/learning-to-rank/): The system is trained on the behaviour it caused.

[Snippet generation](https://www.searchresultoptimization.com/information-retrieval/snippet-generation/): It is written by a machine, from your text, for a query you did not know about.

[Query expansion](https://www.searchresultoptimization.com/information-retrieval/query-expansion/): You are answering a question that was quietly edited before it reached you.

[Relevance feedback](https://www.searchresultoptimization.com/information-retrieval/relevance-feedback/): Absence produces no signal.

[Query performance prediction](https://www.searchresultoptimization.com/information-retrieval/query-performance-prediction/): A page of ten results looks the same whether the answer exists or not.

[Passage retrieval](https://www.searchresultoptimization.com/information-retrieval/passage-retrieval/): Your page is being read in fragments by a system that never sees the whole.

[Mean average precision](https://www.searchresultoptimization.com/information-retrieval/mean-average-precision/): A score of 0.87 tells you how well the ranking matched a list somebody wrote down.

[Vector space model](https://www.searchresultoptimization.com/information-retrieval/vector-space-model/): If the web never described your case, the vector for it is somebody else's.

[Inverted index](https://www.searchresultoptimization.com/information-retrieval/inverted-index/): The web is the part of knowledge that somebody bothered to publish and a crawler managed to reach.

[Precision and recall](https://www.searchresultoptimization.com/information-retrieval/precision-and-recall/): The two measures behind every search evaluation, what they cannot see, and who decided what counts as relevant.

[Latent semantic analysis (LSA)](https://www.searchresultoptimization.com/information-retrieval/latent-semantic-analysis-lsa/): Latent semantic analysis builds meaning from nothing but word counts.

[Controlled vocabulary](https://www.searchresultoptimization.com/information-retrieval/controlled-vocabulary/): A controlled vocabulary fixes which word stands for which concept.

These pages state what each mechanism does, and then the part that matters to anyone who publishes on the web.

![A blurred search engine result page](https://www.searchresultoptimization.com/images/search-result-serp.webp?v=1648967210)
*A result page presents a fraction of what exists. The fraction is the product.*

## The other fields
<https://www.searchresultoptimization.com/information-retrieval/#the-other-fields>

- [Digital marketing](https://www.searchresultoptimization.com/digital-marketing/)
- [Human-computer interaction](https://www.searchresultoptimization.com/human-computer-interaction/)
- [Computational linguistics](https://www.searchresultoptimization.com/computational-linguistics/)
- [Social epistemology](https://www.searchresultoptimization.com/social-epistemology/)

## Everything in this field, on one screen

Each point is a page of information retrieval. Drag, zoom, and click a point to open that page. The whole site is drawn on the [knowledge graph](https://www.searchresultoptimization.com/knowledgegraph/).

[See this field inside the whole site →](https://www.searchresultoptimization.com/knowledgegraph/)
