B-Course is a free web-based data analysis tool for Bayesian modelling, in particular dependence and classification modeling. It can also be used as an interactive tutorial which provides you with data sets that have been prepared in advance.
A game based on English Wikipedia. Your goal is to navigate to a target wiki page by selecting similar pages.
"Search without words". Search Wikipedia pages based on their textual content: define "query" as a set of pages (=combined textual content of the selected pages).
Search Finnish Wikipedia, maximally simple interface. Type something, refine query results by thumb up/down.
Search a subset of Finnish metal industry standards by pasting example text. Visualize search results and do visualization-based query refinement.
Search Buffy the Vampire Slayer episodes using subtitles and transcripts aligned with video. Combines text and image similarity. (Requires password.)
Search Buffy the Vampire Slayer episodes using subtitles and transcripts aligned with video. (Requires password.)
Old, unsupported demos
Search Finnish Wikipedia, control ranking by providing relevancy context by cue words
("Disease Oracle"). Search medical patient records based on both their textual content (free-format case description) and the structured data contained in the form. More information: Löfström, Jaakko: Tiedonhaku sairaskertomusdatasta. M.Sc. thesis, Report C-2008-199, Department of Computer Science, University of Helsinki.
Some early attempts towards adaptive, personalized, collaborative learning environments.
Some early GSM positioning trials, performed on Manhattan in New York. The technology was later commercialized by Ekahau Inc. with a focus in WLAN/indoor positioning.
Examples of 3D visualizations of high-dimensional data. The technology was later commercialized by BayesIT Inc. as the BayMiner web service. It has also been used in several "election candidate selection machines" ("vaalikone" in Finnish), for example in the visualization tool offered by Helsingin Sanomat, the main Finnish newspaper, in the national communal election in 2008.
BAYDA is a software package implementing Bayesian predictive discriminant analysis, where the aim is to build a model for predicting the value of one discrete (class, group, category) variable using other variables. Although the interface of BAYDA was designed having social scientists in mind, the software was also used for classification tasks in many other fields related to machine learning and data mining.
Last updated on 22 Oct 2012 by Antti Tuominen - Page created on 25 Sep 2012 by Petri Myllymäki