This update covers a wide spectrum of Jotup features and is intended to bring most of the system backend to production-ready status.
- search UI:
multiple search types added in the search area, incl. for news, organisations, own content and tags
search boxes re-arranged
explanations added for each field to explain them to the user
'save to Newsfeed' block to save a search as a regularly-running Newsfeed in a Notebook
in-text string search
autocomplete is now much faster and more precise
- added AI-powered 'Tag Suggester' block to automatically suggest tags to the user
- Aggregate news feeds (e.g. on 'My Activity & News' page) now show which individual Newsfeed the news item came from
- Added 'add to favourites' and 'save to notebook' buttons to news items displayed in Newsfeeds
- Clicking the button again removes the news item from the favourites or the chosen Notebook location, respectively
- AI-based Tag Suggester block and new AJAX-based Newsfeed creation procedure to guide the user in efficiently creating a feed
⚡ Summarisation / Insight
Most tag view pages now feature a 'frequency' chart showing frequency of news / content posted within each week/month. These visualise the distribution of events pertaining to the tag within the time dimension. For example, for Organisation or Stock Symbol tags, a frequency chart provides instant insights about the organisation's activity over time. For Locations tags, the frequency chart could reveal spikes of noteworthy events, or seasonal patterns.
- Geotags are now all in uppercase and support UTF-8 encoding for better searching
- Fixed a bug which allocated the wrong vocabulary ID to some geotags, rendering them unusable
- Updated to reflect new concepts of spring of 2023 (over 50,000 new tags, including ChatGPT, Omicron COVID variant, etc.)
- 'Related Tags' block with suggestions for linked (although not synonymous, as Jotup always uses 1 main tag for all synonyms) tags
- summary length is now dependent on the original (longer original = slightly longer summary)
- fixed issue with repeating sentences at the end of the summary if the summariser runs out of things to say
Major update improving accuracy of news summarised by the platform, as well as adding new features.
- 📣 News can now be generated from web pages, rather than just RSS feeds (which are increasingly on their way out) and various XML news formats the system already receives news from. This enables Jotup to scan many new sources, including many previously uncovered organisations
- the system automatically determines news page structure and which elements to scan using user inputs
- it will soon be possible for the user to set up their own news feeds, by scanning either RSS or web pages
- Organisation names are now tags, which can be clicked, leading to organisation profile (see separate update regarding the Organisation and Stock Symbol tag taxonomies)
- Improved the way Location tags are generated - now, in addition to locations supplied by the content provider in via metadata, Jotup also finds locations within the article text, leading to a more comprehensive geotagging
- Where necessary, Jotup can now generate long-form summaries reflecting article structure, just like for Web Summs
Web Summ content type
The Web (Page) Summary content type has been overhauled to be very reliable.
- It can now summarise a wider scope of pages, including many pages under crawl protection.
- In case of failure to either retrieve or summarise a page, it will give better errors.
- Better handling of various encodings and formats
🏢 Lead content type
The 🏢 Lead content type is now self-summarising, just like Web Summ or News content types. The autosummariser can be triggered in 2 ways by the user:
- Inserting a URL in lieu of title during saving will crawl the URL and generate a Lead based on the organisation that is at that URL
- Typing the organisation name into the title field will cause the Lead to be filled in automatically if Jotup already knows that organisation
In either case, Jotup will create a 🏢 Lead profile with the following data:
- description of the organisation
- link to main Organisation page via Organisation tags, if Jotup identifies the organisation as one already listed on Jotup
- any relevant news pertaining to the organisation, listed in a side panel
- stock data and basic financials
- tags pertaining to what the organisation does and locations it operates in
Creating 🏢 Leads and/or 👤 Contacts within Notebooks will enable a Pipeline tab view, which allows the user to interact with the 🏢 Leads, 👤 Contacts, 📅 Events and any related nodes within a more conventional CRM-style view which allows sorting and filtering. There's also a separate People overview tab.
The content type has its own triple backslash (\\\) commands, which have been listed in the \\\ helper block in Outline Designer.
📑 Outline Designer
- Various minor cosmetic changes and bug fixes for Outline Designer
- Fixed buggy move mechanism which previous caused the dragged item to fall into closed branches it was being dragged past
- Added spinning 'in progress' indicator when any manipulation with content is done within Notebooks in Outline view
- Operations taking a long time will additionally display a reassuring 'process still in progress' displaying various live data about the process and server in order to provide reassurance
- Outline Designer can now handle dropped network connections - rather than silently timing out as before, it warns the user about lost connection to the server via visual indicator and resumes the save operation once the connection is re-established. This is useful e.g. when working in a moving setting like a train, or with a mobile internet with poor reception that intermittently drops out.
- Fixed certain Notebook tabs not showing in OD view
- Abstractive text summariser is now a stand-alone server process
- Text summarisers and taggers now have vastly improved request queues, allowing for handling multiple simultaneous requests
- Fonts improved throughout the system.
- Various security fixes & updates to the entire stack
- Completely overhauled antibot / anti-DDOS system, integrating it with leading cloud-based proxy services, resulting in a dramatic reduction in bot traffic and consequent dramatic increase in the amount of simultaneous legitimate visitors the system can support on existing hardware. Suspicious users will have to pass extra AI-based checks rather than getting banned outright as before.