November 22nd, 2011
· This version introduces a RESTful API, the ability to post results directly on the P2P collaborative search ring, the ability to cross-post recommendations to public nodes, and many bugfixes and improvements.
September 14th, 2011
· This version fixes a couple of memory leaks and improves some error logs.
· Overall, it is more stable and production ready.
August 30th, 2011
· This version brings many bug fixes, as well as new feed parsers (for Delicious, Wordpress, Redmine).
· The most notable improvement is a speed-up of P2P search for long queries.
July 3rd, 2011
· This is the first release with peer-to-peer activated by default.
· Every Seeks node defines a list of Seeks peers to work with.
May 2nd, 2011
· Reworked parser and feeds / search engines connection. Allows using the same parser over several feeds. Each feed can be accessed independently from the API.
· Added support for the following result formats: opensearch (ATOM & RSS), dokuwiki and mediawiki.
· Slightly improved personalization over halos of similar queries.
· Few bug fixes, on personalization and on parsers.
· User db optimization is now optional.
· Improved exception model and handling in all plugins.
· Added info API call to HTTP server plugin.
· Many fixes to JSON search output.
· Support for remote user database, for personalization and other means. This allows to deport the user database away from the Seeks node itself. It is the first step towards enabling of a mixture of remote user profiles, in forthcoming version.
· Personalization using remote user database.
· Added new user db remote service plugin (udb-service) to serve user database records remotely, over HTTP.
· Added support for Tokyo Tyrant.
February 14th, 2011
· Added active rejection of personalized results. This allows a negative user feedback that is taken into account in ranking and recommendation.
· Added automated recommendation of queries, URLs (more results to a query). Uses an improved query-to-query weighted distance.
· Added support for remote user database with Tokyo Tyrant.
· Several bug fixes, from language detection to missing results from search engine parsers.
· Fixes to image search.
· Added detection / handling of non-UTF-8 strings for queries and URIs. (includes optional support for libICU).
· non-UTF-8 user db entry fix runs once at first startup of this new release.
· Fixed service of opensearch.xml
· Fixed minor leaks.
· Cleaner destruction of objects at node shutdown.
· Fixed encoding with Blekko parser.
· More unit tests.
December 6th, 2010
· This version adds a new "compact" theme as the default search interface.
· A framework for themes now allows contributors to create new themes with no dependency on the inner search machinery.
· This release contains a dynamic JSON-based user interface with basic features.
· It is faster to use and ready to be plugged in to the P2P protocol when it is released.
September 21st, 2010
· This release introduces microblogging and video search, with results from Twitter and Identi.ca for the former, and from YouTube and Dailymotion for the latter.
· This includes automatic aggregation of retweets and automated clustering of tweets and videos (on the basis of their text title and/or surrounding texts depending on activated options).
· Yauba was added as an additional text engine.
· There were also fixes and improvements to the REST API.
August 19th, 2010
· This version brings image metasearch in the form of a new plugin as well as many improvements to memory management and stability and a reworked UI.
· The Image metasearch plugin integrates results from Bing, Yahoo!, Google, Flickr, and Wikimedia Commons, and embeds an image similarity analysis tool based on OpenCV.
June 11th, 2010
· This releases brings two plugins, an embedded light HTTP server for easing the deployment or personal and public search nodes; a URL blocker for controlling trafic through the proxy (e.g. ad-blocking). The release has an improved overall stability.
· Light embedded HTTP server plugin based on libevent. Support for direct websearch, opensearch.xml and robots.txt.
· URL blocker plugin for controlling trafic through the proxy, such as ad-blocking and more, based on a (by default empty) list of positive and negative regexps.
· Per-query selection of search engines with parameter "engines=" in query URI.
· Improved stability with the correction of some rare but critical bugs and and a memory leak.
· Fixed parsing of some engines results.
· Fixed swedish stop word list.
· Fixed bug when loading stopwordlist at startup for system-wide installation.