FrameNet
The FrameNet project is building a
semantically-rich lexicon of English and a corresponding set of
annotated texts, based on more than 600 semantic frames and 130,000
sentences. Comparable FrameNet projects are underway for Spanish, German,
and other languages. By providing a layered semantic representation of
text, FrameNet delivers a key component of next-generation question
answering, machine translation, and other natural language processing
applications. The FrameNet Website >>
AQUAINT
Researchers in ICSI's AI Group are participating in a project to study deep inferencing techniques and corpus based techniques for deriving the conceptual semantics needed to achieve this. This research is a collaboration with Stanford and the University of Texas at Dallas, and is sponsored by the ARDA AQUAINT Program. Our effort is being intergrated into an ambitious overall program to significantly advance the automated analysis of information.
Semantic Web Services
The Semantic Web is an exciting vision for the evolution of the World
Wide Web. Adding semantics enables structured information to be
interpreted unambiguously. Precise interpretation is a necessary
prerequisite for automatic Web search, discovery and use. Services are
a particularly important component of the Semantic Web. A semantic
service description language can enable a qualitative advance in the
quality and quantity of e-commerce transactions on the Web. The OWL
Services Coalition under the guise
of OWL-S, has taken some important first steps in this direction.
The model of actions, processes and events developed within the NTL
project provides a natural, distributed operational semantics that may
be used for simulation, validation, verification, automated
composition and enactment of OWL-S-described Web services.
NTL
The NTL project of the AI group works in collaboration with other units on the UCB campus and elsewhere. It combines basic research in
several disciplines with applications to natural language processing
systems. Basic efforts include studies in the computational,
linguistic, neurobiological and cognitive bases for language and
thought and continues to yield a variety of theoretical and practical
findings.
One ongoing applied effort (called EDU for Even Deeper Understanding)
has been in operation since July 2000, with multi-year funding from
the Klaus Tschira Foundation. A major aspect of this collaboration has been the interational workshops on Scalable Natural Language Understanding Systems (SCANALU).
The NTL
Website >>
Episodic Memory System
ICSI researchers, working with funding from DARPA's Perceptive Assistant that Learns program, are developing an
episodic memory system for an enduring,
personalized cognitive assistant. The goal of the DARPA project is to
develop an enduring, personalized cognitive assistant in the form of a
software system that can reason, learn from experience, follow
instructions, explain its actions, reflect on its experience, and
behave robustly in unexpected situations (CALO).
SHRUTI: From Simple Associations to Systematic Reasoning
Humans are capable of drawing a variety of inferences effortlessly, spontaneously, and with remarkable efficiency --- as though these inferences are a reflex response of our cognitive apparatus. This remarkable human ability poses a challenge for cognitive science and computational neuroscience: How can a system of slow neuron-like elements represent a large body of systematic knowledge and perform a wide range of inferences with such speed?
SHRUTI attempts to address this challenge by demonstrating how a connectionist network can encode a large body of semantic and episodic facts, systematic rule-like mappings, knowledge about entities, and types, and yet perform a wide range of reflexive inferences within a few hundred milliseconds. SHRUTI Webpage >>
SMRITI: Episodic Memory Formation via Cortico-Hippocampal Interactions
The ability to remember events in daily life and acquire specific facts after reading a newspaper or watching a newscast underscores our ability to rapidly acquire new memories. In general, these memories encode who did what to whom where and when, and have been described as episodic memories. Neuropsychological, anatomical, and physiological data suggests that the hippocampal system (HS) plays a critical role in the encoding and retrieval of such memories. But how the HS subserves this mnemonic function is not fully understood.
SMRITI is a computational model that demonstrates how a cortically expressed transient pattern of rhythmic activity representing an event or a situation can be rapidly transformed into a persistent and robust memory trace as a result of long-term potentiation and long-term depression within structures whose architecture and circuitry are consistent with those of the HS. SMRITI Webpate >>
Color, Language, and Thought
In 1978 The World Color Survey collected color naming data in 110 unwritten languages from around the world. The ICSI WCS staff (Paul Kay and Richard Cook of ICSI, Terry Regier of U. of Chicago) has recently put these data into a single data base, available to the scientific community. Several outside laboratories have already used this data base for studies.
ICSI WCS personnel, joined in one case by Michael Webster (U. Nevada, Reno), have recently published three statistical studies, using the WCS database, that establish universal tendencies in color naming (contrary to the claims of some linguistic relativists).
A collaboration of Aubrey Gilbert and Richard Ivry of U.C.B. Psychology with Regier and Kay has demonstrated Whorfian effects in color discrimination lateralized to the right visual field. The right visual field projects to the left cerebral (or "language") hemisphere. A follow-up study is nearing completion in collaboration with researchers at the University of Surrey led by I.R.L. Davies. Future lateralization studies of color, other lexical domains, and other languages are planned.
Regier, Kay and Naveen Khetarpal (U. of Chicago) are completing a study, based on the WCS database, predicting universal tendencies in color naming from the interaction of a general principle of category formation with the surface of the Munsell color solid (represented in the CIE L*a*b system). Further studies in this line are planned, using different color order systems and more complex and empirically realistic models of the evolution of color naming systems.
More about the AI
Research Group >>
top |
