Replies: 2 comments 1 reply
|
@fschade Could we implement something like that via our OpenSearch backend? |
0 replies
|
fuzziness is not a problem! I have a POC running here that uses dGraph as a vector store and enables this kind of thing... but theoretically, it could also be done using standard OpenSearch features.... but as usual it depends what we wanna do, we have to pick the tools depending on that... if we wanna have relational matches weaviate would be a good fit too... |
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment


Uh oh!
There was an error while loading. Please reload this page.
I am referring to my introduction to this post, in which I had already mentioned that I had tested quite a few FileSync cloud solutions. So here is my idea for the search.
https://github.com/orgs/opencloud-eu/discussions/3126
Summary:
It would be a unique selling point if the web and app versions were still able to find search terms that were misspelt or words in PDF scans that had been misrecognised.
EN:
We have personally digitised around 5,000 pages or more in PDF format, covering the period from 1992 to the present day. Our files are currently synchronised on our Mac via Google Drive. The full-text search in Google Drive works better than in OneDrive, both in the app and in the web version, provided the word appears in full within the PDF or document. (Zero-error tolerance)
On a Mac, the search works almost flawlessly, provided the documents have been processed with OCR using Abbyy FineReader. The recognition is so good that even text fields – whether typed or in scanned documents – hardly matter anymore.
To date, there is no web, app or cloud file-syncing solution that can reliably handle text recognition errors, imprecise searches or fuzzy searches.
It would therefore be great if the OpenCloud app and web version supported fuzzy search. This would be a real unique selling point for OpenCloud, as other cloud file-syncing services do not currently offer this feature.
Provided that the PDFs contain OCR data, the Finder search on the Mac is currently unbeatable and indispensable. It finds practically everything – regardless of how incorrectly a word has been recognised in the scanned document.
At present, however, no app or web solution offers this level of convenience when you’re on the move.
OpenCloud apparently offers a powerful full-text search even for non-OCR PDF documents. The only thing missing now is the ability to handle misrecognised and misspelled words.
DE:
Wir haben privat etwa 5.000 Seiten oder mehr aus dem Zeitraum von 1992 bis heute via PDF digitalisiert. Unsere Dateien werden aktuell über Google Drive auf dem Mac synchronisiert. Die Volltextsuche funktioniert in Google Drive sowohl in der App als auch in der Web-Version besser als bei OneDrive. Sofern das Wort vollständig im PDF / Schreiben angegeben wurde. (0 Fehler Toleranz)
Auf dem Mac funktioniert die Suche nahezu einwandfrei, sofern die Dokumente mit OCR durch Abbyy FineReader verarbeitet wurden. Die Erkennung ist so gut, dass selbst Textfelder beim Tippen oder in eingescannten Dokumenten kaum noch eine Rolle spielen.
Bisher gibt es keine Web-, App- oder Cloud-File-Sync-Lösung, die zuverlässig mit Texterkennungsfehlern, unscharfer Suche oder Fuzzy Search umgehen kann.
Es wäre daher großartig, wenn die App und die Web-Version von OpenCloud eine unscharfe Suche bzw. Fuzzy Search unterstützen würden. Dies wäre ein echtes Alleinstellungsmerkmal von OpenCloud, da andere File-Sync-Cloud-Dienste diese Funktion derzeit nicht bieten.
Vorausgesetzt, dass OCR-Daten in den PDFs enthalten sind, ist die Finder-Suche auf dem Mac aktuell unschlagbar und unverzichtbar. Sie findet praktisch alles – unabhängig davon, wie fehlerhaft ein Wort im eingescannten Dokument erkannt wurde.
Derzeit bietet jedoch keine App- oder Web-Lösung diesen Komfort für unterwegs.
OpenCloud bietet ja scheinbar eine starke Volltextsuche auch bei nicht OCR PDF Dokumenten. Jetzt fehlt zu dem Thema nur noch das es mit Falsch erkannten und falsch geschriebenen Wörtern umgehen kann.
All reactions