Swirl Metasearch 2.5.1
Team Swirl is announcing General Availability of Swirl 2.5.1!
This release features three new SearchProviders (arXiv.org, EuropePMC.org, and LinkedIn profiles via Google PSE), enhanced result processing to handle XML responses from source APIs, plus updates to the ChatGPT Connector
and QueryProcessor
to support the latest GPT-3.5-Turbo and GPT-4 models.
🥇 We would also like to give a shout out to GitHub user Rajpratik71 for his contribution to this release which improved our Dockerfile
configuration. Many thanks for contributing to Swirl!
PLEASE STAR OUR REPO: https://github.com/swirlai/swirl-search
New Features
🔹 Both the ChatGPT Connector
and QueryProcessor
were updated to use OpenAI's ChatCompletion
method which supports the latest GPT models - including GPT-4 - and a much greater range of interactivity.
- Three new Tags and one new
query_mapping
option are available for the ChatGPT SearchProvider to help shape the Prompt or Default Role passed to ChatGPT along with the user's query. - The ChatGPT SearchProvder now queries the GPT-3.5-Turbo model by default.
🔹 Search LinkedIn profiles with the new LinkedIn Google PSE and SearchProvider.
🔹 Search the academic paper collections of arXiv.org and EuropePMC.org with new SearchProivders for each.
🔹 New result processing to handle XML responses from source APIs; see this in action with the arXiv.org SearchProvider.
🔹 Swirl's Dockerfile
now uses the --no-cache-dir
flag with pip
to save space.
Thanks to GitHub user Rajpratik71 for this contribution!
Improvements
🔹 The Subscribe functionality was reworked to better handle Microsoft Authentication for subscribe runs that include M365 sources.
🔹 Some default SearchProviders have additional Tags: Internal
or Develop
. See the documentation for details about targeting specific groups of sources using SearchProvider Tags.
🔹 The install.sh
script was further enhanced to synchronize spaCy model and code version updates.
Resolved Issues
🔹 Addressed an issue where Swirl could hang due to very long source repsonse times. Swirl's default timeout value in swirl_server/settings.py
is now: SWIRL_TIMEOUT_DEFAULT = 10
🔹 Corrected a count inconsistency in the CosineRelevancyResultProcessor
🔹 Removed duplicate entries from urls.py
🔹 Fixed a regression where unmatched Tags were being dropped in searches
🔹 ChatGPT no longer returns a 500 error when enabled for Subscribed searches
🔹 After authenticating to Mircosoft in the Galaxy UI, the first search run now correctly includes M365 resutls
🔹 Addressed an issue with date-sorting in the M365 sources:
- The default date-sort for M365 sources is:
DEFAULT_DATESORT_X = "createdDateTime desc"
- This can be overridden by setting
DATE_SORT
in the SearchProviderquery_mapping
- Owners of a Search object or Swirl's Subscribe feature can set the
sort
attribute on a Search
Known Issues
🔹 Clicking a Microsoft Teams result sometimes results in this error from the Teams app:
We cannot take you to that message because it's in a chat you're not in.
For Teams links to resolve correctly, you must have the Microsoft Teams app open before clicking the result link.
🔹 Creating searches from a browser with q= can sometimes create two Search objects.
This is because of browser prefetch AKA predictive service. Turn off Chrome prediction service. Turn off Safari prefetch.
Please report any issues with this to support.
Upgrading
Documentation
Overview | Quick Start | User Guide | Admin Guide | M365 Guide | Developer Guide | Developer Reference
(updated links on 13-October-2023)
Support
🔹 Join the Swirl Metasearch Community on Slack!
🔹 Email: [email protected] with issues, requests, questions, etc - we'd love to hear from you!