You know more about a file than its inventory says: file notes

Anyone who has worked with a file knows more about it than its inventory says. Which villages and years it really covers. That the archive got the title or the years wrong. That this book is a copy of another one, from the consistory's fond. How a surname is really written when at first glance it is easy to read it differently.

Until now this stayed in your notebooks and in your memory. The next researcher who opened the same file started from scratch, and the machine text misled them in exactly the same way.

Now you can add notes to any file. Your knowledge of the file stands next to the text the program read, and it carries more weight: the program guesses, but you have seen the document.

What you can write down

• What the file really contains. Villages, years, parishes, genre — the way you know the file, not the way the inventory named it.
• A mistake in the archive's inventory. The inventory says 1834, but the book covers 1834–1836. The title says “parish register”, but inside there are confession lists.
• A copy or a related file. A fair copy in the consistory, a draft in the parish fond, the continuation in the next inventory.
• How it is really written. A difficult surname, first name or date you have worked out — exactly as it appears in the document.
• Your own, family notes. “My grandfather is recorded here” — for yourself. Nobody but you sees such an entry.

Who sees it

Supriaha will not accept an entry containing “my grandfather”, “our family”, “great-grandmother …”, a link to a person from your tree, an email address or a phone number. Write your knowledge of the file as a separate entry for everyone, and family matters as a separate one, for yourself. The literal text of a document is not limited by this check: “правнук его Иван” or “Отец наш” in a revision list are the words of the source, and they pass.

A moderator can hide an entry from the book's page. After an account is deleted, its entries remain as knowledge about the file, but without a signature. An entry you withdraw disappears from the book's page together with its clipping.

Your reading teaches Nyshporka

The places you have read correctly are exactly the ones where the program went wrong. That makes them the best material for teaching it to make out old handwriting better.

• Your readings can be used to train your own model on your computer.
• When you share a reading in Supriaha, a clipping of that place from the scan goes with it (if the file's scans are on your computer). By sharing readings, you agree that the owner of Nyshporka may use them to train recognition models. Clippings are not published and are not given to other users.

How to add a note

On the website. Open the book's page in Supriaha, sign in and click “Add an entry” in the “File notebook” section. This way you can add a file description, a mistake in the archive's inventory or a copy — you do not need to install Nyshporka for this.

In Nyshporka. Ask your assistant: “add to file ДАВіО 904-24-54 that it is a copy of the book from the consistory”, “write down how the surname in entry 153 really reads”. The assistant knows the commands itself. For those who work without an assistant:

Questions and answers

How are notes different from the file text read by Nyshporka?

The program reads the whole file from start to finish and makes mistakes here and there. Notes are written by you, about what you know for certain. Search shows your reading next to the machine line, so next time the surname will be found even if the program garbled it. But notes do not count as a read file: if you wrote down three places, that does not make the file “read”, and the conclusion “this surname is not in the file” does not rely on them.

Can I share a file description without handing over clippings for training?

Yes. File descriptions, inventory mistakes and copies have no clippings, so you can share them freely. A line reading goes together with a clipping; if you do not want to hand it over, keep the reading for yourself — it works just the same for search on your computer.

How do I remove an entry I have already shared?

Ask your assistant to withdraw the entry, or run nysh note retract <справа> <ідентифікатор> and nysh note push <справа>. The entry will disappear from the book's page, and its clipping will be deleted.

I already have my own notes on files in folders. How do I move them?

Folders where lines are written as [рядок N] … are moved by nysh note import-eye <тека>. Readings become entries on the file, remarks become your personal entries; what to share in Supriaha is up to you.

How can I see what others have added to a file?

On the book's page in Supriaha, or with the command nysh note pull <справа>. Other people's entries are added to your file read-only, marked with who added them.

What else is new in this update

Searching among victims of repression

Nyshporka searches the “Ukrainian Martyrology of the 20th Century” database of the State Archival Service — a name index of people repressed in the 1920s–1950s. By surname (and first name, if known) it shows dates of birth and death, where the person lived at the time of arrest, the sentence, rehabilitation and, most importantly, the number of the archival investigation file — the archive, fond, inventory and file you use to order it from the archive.

Ask your assistant to “search for Ivan Petrenko among victims of repression” or run nysh find "Петренко Іван" --source martyrolog. The database website is protected against automated requests, so this search needs a Nyshporka add-on; if Nyshporka says it is missing, ask your assistant to install it.

The database covers only some of those repressed, and unevenly: from one region there are hundreds of people, from another tens of thousands. Under every search, Nyshporka writes how many people from each archive are in the database — so you can see what a zero is worth in your particular region. If a person is not found, it is no proof that they were not repressed; the database keeps growing, so it is worth repeating the search. The general search “what is there about my village” does not query this database: it is about people, not villages.

A village's parishes: more names, more books

• Nyshporka tries other spellings of the village itself. The archives index finds a parish only by the spelling it was recorded under: “Мястковка” was not found for “М'ястківка”. Now the Russian and Polish names of the village and its former name are taken from reference works, if they are there. If there are several villages with that name, former names are not used, so as not to end up in someone else's village.
• Alongside — the 18th-century parish from Szady's database. Uniate and Catholic parishes of ~1772 are shown under their Polish name with the deanery, and Nyshporka suggests how to open the parish's card and find its books.
• Large parishes are no longer cut off at 50 books. The index returns at most 50 files at a time; Nyshporka fetches the rest by year, and if that still does not fit, it says so.
• The Ridni catalogue finds books when the county is written in a different case: “Острозький” and “Острозького” now give the same result.

Lists of an inventory's files from Wikisource

The “Archive” project on Ukrainian Wikisource keeps lists of files for many inventories — with titles, years and links to FamilySearch films and Commons files. Nyshporka now collects such a list itself: ask your assistant to “collect the list of files for ДАЖО fond 1, inventory 78 from Wikisource”. What is collected goes into the inventory's register, so you can see which files of the inventory are online and exactly where. Sometimes the combined index links only some of the yearly volumes to a parish, while the full series is here.

Searching what has been read

• “Search” sits in “Basics”, right after “Overview”, and has two fields. “What we are looking for” suggests the spellings of your family name and recent searches, but you can also type a village or any word. “Where” suggests the fonds, inventories and files that have something read: “ДАВіО fond 904 · 50 files”. Nyshporka takes the family name from your profile itself, and within a single file it also looks for the family's first names nearby and for signs of a family in one entry. The app searches the same way as the assistant, so your answer and its answer do not differ.
• Under the finds — “Where and how we searched”: how much of what has been read was combed through, and what was not checked is named, with the reason. So you can see what a zero is worth.
• The real surname comes first. Previously, every find where the search term appeared even inside a longer word had a similarity of 100. Now 100 means only an exact match, each line says what kind of match it is (“exact”, “similar word”, “different ending”, “inside a longer word”), and exact matches come first. Nothing has been thrown out: a first name and surname run together can also be your family, so they stay at the bottom.
• A whole inventory or fond in one search. Name an inventory — for example, “ІА 630-1” — and the search will go through all the files that have been read, and tell you as a number how many have not been read yet. A fond of a hundred files is searched in half a minute instead of two and a half, and under the field you can see where the search is: “files of series 37 of 109”.
• Search does not freeze or get lost. You can see exactly what it is doing; “Stop” really stops it; if you leave the screen and come back, the search will still be there.
• An honest search status. Nyshporka no longer says “thousands of runs outside search” when in fact two are missing. “Build index” catches up only on what is missing, and under the button you can see how long is left to wait.

Searching the catalogues

• Five times faster. All sources are queried at once, and under the field it says how many have replied and who we are still waiting for (“waiting for: Internet Archive”). The search can be stopped — then you see the finds from the sources that managed to reply.
• An incomplete catalogue is called just that. If collecting a source's catalogue was stopped or interrupted, the list of sources says “crawl not finished”, and the search warns that finds from there are incomplete. A catalogue is collected with the “Build catalogue” button in the list of sources, and pressing it again continues rather than starting over.
• From a village's card straight to the catalogues. The “Catalogues →” button on a village's card opens a search with its name, and next to each file you can see whose archive it is.

Reading: pause, change, finish

• “Stop” really stops reading. Previously, reading could be marked “stopped” but in fact carry on and keep the next file away from the graphics card. Now everything stops, and if you close Nyshporka, reading closes with it.
• Reading progress — on the file's card: how many pages have been done, whether reading is waiting for the graphics card, and a “Stop” button. While a file is being read, it cannot be started again.
• “Change” pauses reading and opens the same form: you can switch the script or add a voice, and what has already been read is not lost.
• “Add a voice” on the card of a file that has been read: a second engine reads all the pages separately, overwriting nothing.
• You can see what has not been finished. In “Overview” and in “Runs”, every reading is marked: “done”, “reading now” or “not finished”. “Finish reading” continues from where it stopped. If all pages already have text, the plan says so — instead of starting something that would do nothing.
• Starting is instant. “Read” no longer waits 15 seconds at a time: you immediately see “Adding to the queue…”, and a second later “Reading now”.

PDF files taken with a phone are read in full

Pages taken with a phone are often stored sideways in a PDF and only appear upright when viewed. Nyshporka used to send them for reading sideways, and only a few lines came out of a folio instead of all of them. Now a page is read the way you see it. Pages that were previously read sideways Nyshporka will rotate itself as soon as you open reading of that file again, and it will say how many there were. It will set aside what was read from them, so “Read” will finish exactly these pages and leave the rest of the file alone.

No repeat downloads

When taking files from Supriaha, Nyshporka first compares them with what has already been read on your computer and says: this many are already there, this many may be there, this many are missing. It takes only what is missing. A file posted on Wikimedia Commons as several files (volumes) is downloaded in full, not just the largest volume.

The app is faster and calmer

• The home screen opens at once instead of showing grey bars for 10–15 seconds; the “File queue” no longer slows down with every file added.
• Only what really needs fixing, so that reading or search do not give a wrong result, is highlighted in yellow. The rest of the explanations are folded into a “ℹ Notes” line.
• The library counts files, not folders; files whose scans were removed to save space are shown separately — “removed from disk”. Click a file's row — its full title, folder and actions appear below it: view folios, read, search for the family.
• “Reception” is now called “New folders” and shows only what is waiting to be described, grouped by origin: a hundred issues of one newspaper are one row. Next to each folder — “Folios”, “Describe” and “Not a file”. While you type a reference code, you can see under the field how Nyshporka has read it: archive, fond, inventory, file.
• The main tab is now called “Overview”, so as not to confuse it with the “Research” section.

The assistant knows when to turn to Nyshporka

The assistant chooses a Nyshporka tool by its short description, and when there are many tools on the computer, the descriptions of rarely used ones drop out of its view — precisely the ones that came with an update. Nyshporka's tool descriptions are now half as long, and nysh doctor checks whether the assistant can see them and suggests what to change if not.

Compatibility

• A file with notes added cannot be read by earlier versions of Nyshporka. Update Nyshporka on all computers that open the same folder. Files without notes do not change.
• Notes first appeared in version 0.25; 0.26 added training on checked lines and fixed several faults in notes; 0.27 brings everything described above in the section “What else is new”.

How to update

nysh update — or ask your assistant to update Nyshporka.