This is the company that powers the FT.com mobile software.


http://www.leiki.com/technology.html
http://www.leiki.com/targeting.pdf

Leiki Targeting: The heart of all the Leiki Products

The architecture and implementation of Leiki Targeting is designed to be efficient under heavy traffic in varying hosting environments, while implementations as client applications on mobile phones with limited resources and slow connections are fully functional.
The key design targets have been:
Highly optimised for CPU and memory: ScalabilityApplicable to a wide variety of content types: FlexibilityEasily integrated to a generic portal or content service: Integrability
The Heart of all the Leiki Products
Leiki Targeting is a software engine that supports real-time learning, profiling and personalisation. It is the only technology available that offers adaptive personalisation via messaging (SMS, MMS) interfaces, client-based (Symbian and Java smart phone) implementations and browser interfaces.
Leiki Targeting automatically creates comprehensive profiles of both all the users and the content in the system. The technology is multi-lingual and delivers personalised information from any underlying content database. Profiling is based on detailed ontologies which can be extended to any specific field allowing easy integration of new content types.
Why Leiki Targeting?
Increased usageServices offer content personalised to the user. They add value with no additional user effort required and therefore result in increased usage.
LoyaltyContinuously improving personalisation boosts customer retention.
Market infoStatistical analysis of user interest profiles generates very valuable information about market segments and behavior.
AutomationSignificant cost savings through automation of what were previously manual tasks.
Semantic Web compatible
Leiki Targeting is compatible with the W3C standards RDF, RDF Schema and OWL. Ontologies developed according these standards can be easily imported to the Leiki system for editing with our advanced Ontology Editor and use in content services ranging from SMS matchmaking to virtual personalities running on Symbian smartphones. Likewise, ontologies in existing Leiki services (such as a 10k+ category IPTC-based General Ontology) are easily output for other systems that use Semantic Web standards.
Technology Highlights
Natural language understanding. Specific parameters are not needed in the user input.
Messaging user interfaces. Leiki is unique in providing adaptive personalisation via the mobile messaging (SMS, MMS) interfaces.
Client-based implementations (Symbian & Java smart phone, PocketPC and PDAs). The client-based implementations have important privacy advantages: the user profile exists only on the user's client-based phone and only the user can see it. This also enhances the speed and opens new areas for applications.
Focus on Computing Quality
Strong hybrid personalisation method- Probability-based interest level ranking- Rule-based selection- Social filtering
Significant R&D optimisation efforts- Profile size is extremely compact- Computing time for profiling and profile matching is short
Technical Information
Java API, run-time edition 1.3.x or higher preferred, runs on 1.2.x. Installations on Sun UltraSparc/Solaris, Windows/Intel PC
JDBC connection to SQL DB
Installations on Oracle, MySQL, DB2, etc
Processes unstructured text, not limited to XML
Hybrid algorithms, latest academic with in-house flavours
Performance (on an Intel 800MHz)- 100 000 match operations/sec- Content profiling: tens of thousands of items per second