extended Classification Threshold configuration, process unsupported files, re-order pages in Separation
A Note from our Head Product
Hello everyone 👋
We kick off March with some cool new features for the Parashift Platform. But more importantly (what could be more important than new features, you may ask yourselves?), I want to hijack these release notes to tell you how we will go forward with product marketing and generally keep all of you out there up to date regarding new releases and features.
In the past, we (I) have been lacking a bit with public information, regular release notes, explainer videos, knowledge base articles and the like. From now on, I will commit to constantly posting release notes and, where needed, more detailed knowledge base articles to go in-depth into the new stuff.
Clear communication is now more important than ever since we have some massive hot features cooking which we plan to release in the upcoming weeks. (some are only a few days away from release 😉 ) But please go easy on me; it won't be perfect from the start but will get better and better in the upcoming weeks.
So stay tuned for more. If you want to get updated about our newest release, for now, use our newsletter where we inform you once a month what's up with Parashift in general but also reference new releases.
Cheers ✌️
Thilo
PS: If you want to always stay up to date as soon as we release new features leave us your email here: Release Notes mailing list
Video Explainer
🎥 Also check out our release explainer video if you like 🎥
More Classification Threshold Configuration options
🔢 When predicting the document type, our Classification spits out a prediction confidence (in per cent, 0%-100%). You could already configure automated processing; let's say if this confidence is above 95%.
We now added more granular controls to take into consideration the document's language so that you can set one threshold per document type and language.
🌍 This helps, especially if you have little Training Data for specific document languages since, e.g., you are expanding into new countries. You can now make extra sure that these low volume document languages get classified correctly while the vast majority of high volume document languages still get processed automatically.
Process unsupported files
For us to be able to process a document you had to upload it with one of the following supported file types (mime-type):
- PDF (of course 😉 )
- jpg
- png
- tiff
These are great and cover the most common use cases for document Classification & data Extraction 🧠 .
But we now also added the option to upload unsupported file types like Word, Excel or any other common mime-type. While we cannot automatically process these, you are still able to upload them, classify them manually, and extract data manually.
This is super useful if you want to automate a process where 99% of all documents are pdf, jpg, png or tiff. We now have a way to handle the pesky 1% of other files without the need for some custom workflow or custom development on your side; all your files can flow through Parashift.
(🕜 Down the line we of course plan to support more file-types for automated processing, this is only an intermediate step to that goal)
Re-order pages in Separation
🔀 When first developing our Separation feature that lets you automatically or manually separate documents in a batch, we seem to have underestimated the chaos out there since we did not allow users to order pages in batch.
To bring order into your documents, we now have added precisely this, the possibility to, well, order pages in batches and documents 😃
Integrations & API
JobRouter System Activity updated to v1.2.2- Field mapping between Parashift and JobRouter does not require mapping every field anymore, making configuration and testing a lot easier
- Added support for Linux
- Fixed: conversion error for DateTime fields
Fixed
- "Exported" tab only shows 20 documents
- "force done" button skipping fields for export