Artificially generatedSemantic Mining: we dig through reality to make meaning
Data is there. The data gets mined. What we actually want is meaning.
Semantic Mining is the making of meaning from the infinity of the unknown. That is how it looks. The world is, in the end, a large data set. That data set comes into our brain, or it goes from one system to another. We now have neural nets and can, for the first time, really work meaning out of the world without using our own brain.
Semantic Mining is the main subject, and of course it is about making more meaning from the world.
We dig through reality to make meaning. That is not a slogan about tools. That is the work. The infinity of the unknown sits out there as data, as feeds, as pages, as satellite images, as tariffs, as songs, as what happened yesterday. Semantic Mining is the name for the digging that does not end in a table, but in meaning.
The paradigm is new, and nobody uses it the way we mean it. Data mining has been around for decades. The semantic web was a promise. What works now works because neural nets can do the second sort: not just fields into columns, but patterns into meaning.
This is not about data mining. Data is there. The data gets mined. What we actually want is meaning.
Data mining pulls more data out of a pile that already counts as data. Correlations, clusters, features. Useful. Not the goal. Semantic Mining starts one layer earlier and stops one layer later. Earlier: the world is still earth, still unsorted, not yet data. Later: from the sorted pile comes meaning that a human or another system can use, without wading through the earth themselves.
Meaning without using our own brain does not mean nobody thinks anymore. It means the brain is no longer the only sorting machine on the yard. Meaning used to have to go through a skull. Now meaning can also go through a net we hook up to the Laster (the truck).
We drive around in the big data mine of civilization and first collect Erz (ore): raw material, earth, unsorted earth, Datendreck (data dirt).
That is how it lands on our giant Laster, and it has to be sorted. Topic-feed intakes. Several. First the ones we have: scrapers that produce feeds, meaning new data streams that come in every day. That should run largely deterministic. Possibly five hundred scrapers that fire off every day with cron jobs and fetch data. Not because five hundred is a magic number. Because the mine is large, and one intake does not carry civilization.
Datendreck is not a bug in the system. Datendreck is what civilization dumps all day: pages, prices, lyrics, tariffs, planes, oil barrels, communiqués that never arrive. If you only want clean tables, you stay in the wash bay. If you want meaning, you drive into the mine.
It can be pre-sorted with deterministic software. That is what we always did before, and where we still are, mentally, a lot.
Cron. Host. Spacing. One request after another on the same host. Fields that have a name. Date, source, URL, raw text. That is the first sort: from Datendreck comes a crate you can stack. Without that crate the net has nothing to bite but mush.
Deterministic means: the same day, the same source, the same field. No silent fallback that shows your own line when the proxy sticks. No scraper that introduces itself as McGrinsey in the user-agent. The intake layer is boring, and boring is a virtue here. The intelligence sits one layer up.
Now we also have neural nets we can hook this up to, to run smarter, deeper sorting.
The semantic sorting machine is the second device on the yard. It gets the crates from the deterministic layer and turns them into patterns. Not another column. A reading. What belongs to Chinamerica, what to a song, what to a levee that is dry. What is fact, what is hypothesis, what is rumour. What is the heaven in a person, and what sits in the middle without opening the piece.
That is Semantic Mining as a machine, not as a metaphor. The infinity of the unknown stays infinite. The machine does not make it small. It makes it sortable, deep enough that meaning comes out, without a single skull having to run the whole mine.
Topic feeds, daily streams, deterministic. Datendreck onto the Laster. Repeatable.
Fields, sources, date, raw text. Software that already worked yesterday. Mentally we still sit here a lot.
Deeper sorting. Readings, not columns. Meaning, without a brain walking the mine every time.
Sorting is nothing other than the making of patterns, the making of meaning.
A pile of earth is unknown. The same earth, separated by rock, by ore, by dirt, is already a thesis about the world. Semantic Mining takes that seriously: meaning is not an extra you lay on after the table. Meaning is the sorting itself, when it goes deep enough.
That is why two machines, not one. Determinism keeps the Laster at the source. The net keeps it at the meaning. Who only has the first stays in the 2010s, where we still are, mentally, a lot. Who only has the second hallucinates into the mush. Both, in sequence: that is Semantic Mining.
Making more meaning from the world then means, concretely: every day new streams, pre-sorted, then semantically sorted, then a sentence a human can use. An article. A feed. A globe you can spin. The mine stays. The meaning comes out.
| Claim | Basket | From where |
|---|---|---|
| Semantic Mining = making of meaning from the infinity of the unknown | Definition, this page | John McGrinsey, 9 September 2026. Our own term, not an ISO standard. |
| Data mining and Semantic Mining are the same | No | Data mining sorts existing data. Semantic Mining pulls meaning from unsorted world plus nets. |
| Neural nets can work meaning out without a human brain | Hypothesis | This page. Meant: sorting deeper than rules. Not: nobody thinks along anymore. |
| Five hundred scrapers as intake | Architecture sketch, not a built number | Spoken stand 9 September 2026. The number stands for the width of the mine, not for a running cron. |
| Nobody uses Semantic Mining yet | Stance, not a market report | Meant is this reading (meaning from the unknown, two machines). NLP and knowledge discovery exist under other names. |
| Sorting = pattern = meaning | Hypothesis | This page. Meaning as the product of sorting, not as decor after. |
- This textOur term Semantic Mining, mine metaphor, two machines. As of 9 September 2026. German sister: semantic-mining-sinn-unbekanntes.https://mcgrinsey.com/magazin/semantic-mining-meaning-from-unknown/
- ChinamericaExample of a semantic sort: energy and chips first, Taiwan later.https://mcgrinsey.com/magazin/chinamerica-us-china-energy-chips/
- American PieExample: a song as a data pile, meaning as sorting (girl, pie, levee, direction).https://mcgrinsey.com/magazin/american-pie-miss-american-pie-levee/
Research cut: 9 September 2026. Written with the McGrinsey Writing Skill, original tone from the spoken brief. Semantic Mining is our name for the work, not someone else's product name.
