Taking notes in class: the lecturer's PDF, photos of the projector and your notes

A lecture leaves three traces: the PDF the lecturer hands out before or after, the photos you take of the projector, and the notes you type in a hurry while they talk. All three are about the same slide, and nothing ties them together. Mnemosyne OS can bring them back together at the end of the semester, on one condition: they have to share a few words. This page says which ones, and where to put each file.

One vault for your courses

Create a vault with "+ New Vault", call it Courses, and give it a folder to watch with "Add a folder to watch…". Inside that folder, one subfolder per subject. Whatever lands there enters memory: the lecturer's PDF, your photos, your Word files. That is the only filing to do, and you do it once.

Do it before the first lecture, not after. The settings below apply to files that arrive after them, and a file already read is not read again until it changes.

The anchor: the first words of the slide

This is the gesture that makes everything work. When a slide comes up, start your note with its first words, exactly as written, before writing down what the lecturer says about it. The slide's title if it has one, its first line otherwise. Not a paraphrase: the same words.

Why. When you ask a question, memory searches two ways at once, by meaning and by literal words. The lecturer's PDF contains those words. Your note contains them. So does the photo of the slide, once its text has been read. The three passages then come up in the same answer, and the model has them side by side: what the slide said, what the lecturer added, what you wrote down.

Without the anchor, your note says "they stressed the second point" and nothing knows which slide that was. With it, the sentence has an address.

One note per lecture, one heading per slide. That gives you a readable note, and it is exactly the cut memory prefers.

Your notes: in Sovereign Notes, or in Word

The most direct route is the Sovereign Notes window: "New Note", and you type. The note is a file saved in a folder memory already watches, so it enters memory shortly after every save, with nothing else to do. Type the anchor as a heading with ##, and the headings that follow also structure the note's mind map.

If you would rather use Word, that works. Save the .docx file straight into the subject's folder, and its text enters memory at every save. The temporary files Word creates next to it are ignored. The anchor works the same way: the first words of the slide, as a heading or on a line of their own.

One limit to know about with Word: pictures pasted into the document are lost on reading, unless you turn on "Extract the images inside Word documents" in Settings → Images. They then follow the same rules as your photos, below.

Photos of the projector

Put them in the subject's folder, like everything else. Two things to know.

Images do not enter memory by default. Turn on "Enable image memory" in Settings → Images: the switch installs the engine and downloads what it needs. Without it, your photos are seen in the folder and left aside, and a "Not taken in" card tells you so.

The text written on the photo is not read by default either. That takes an OCR engine, installed in Settings → OCR, and the "Read their text" action on that same card. Until that is done, a photo is found by its file name and by visual resemblance, never by the words it shows. And a photo already taken in keeps the text it came in with: turn reading on before you drop the semester's photos.

Formats read: jpg, png, webp, bmp, tiff. The iPhone HEIC format is not read: set the camera to its most compatible format, or export to JPEG before dropping.

A photo of a projector is often blurry or taken at an angle, and reading gets only part of it. That is why the anchor in your note stays the reliable link: that one, you wrote cleanly. If you have a minute, renaming the photo with the slide's first words gives it the same address, even with no text reading at all.

The lecturer's PDF

Drop it in the folder, and that is all. A long document enters whole. A PDF that is only an image, a scan with no selectable text, waits for an OCR engine before entering, and the "Not taken in" card counts it.

Once the lecture is over

Open the chat, set the "Scope" to your Courses vault, and ask. What the lecturer said about a given point, what the slide said exactly, where your notes and the PDF do not tell the same story. The answer names the files it comes from, so you can reopen the photo or the PDF.

Your files never leave your machine. Only the passages found are handed to the model you chose, and a local model sends them nowhere.

If an answer finds nothing although the file is in the folder, the page on empty answers lists the causes in the order to look at them.