BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Women@CL talklet event - Zohreh Shams\; Luana Bulat\; Guo Yu
DTSTART:20170309T130000Z
DTEND:20170309T140000Z
UID:TALK70789@talks.cam.ac.uk
CONTACT:Ekaterina Kochmar
DESCRIPTION:Speaker: Zohreh Shams (Artificial Intelligence group)\n\nTitle
 : Accessible Reasoning with Diagrams\n\nAbstract: Ontologies are notorious
 ly hard to define\, express and reason about. Many tools have been develop
 ed to ease the debugging and the reasoning process with ontologies\, howev
 er they often lack accessibility and formalisation. To address this shortc
 oming\, we propose using “Concept Diagrams” to perform inconsistency a
 nd incoherence checking in ontologies. User studies have proven the cognit
 ive advantages of concept diagrams over Description Logic and OWL (two com
 mon ontology representation languages). With reference two these results w
 e have developed a set of sound inference rules to reason with concept dia
 grams about inconsistency and incoherence in ontologies. We are currently 
 investigating the completeness of the devised set of inference rules paral
 lel to implementing them.\n\n===============\n\nSpeaker: Luana Bulat (NLP 
 group)\n\nTitle: Modelling metaphor with attribute-based semantics\n\nAbst
 ract:\nOne of the key problems in computational metaphor modelling is find
 ing the optimal level of abstraction of semantic representations\, such th
 at these are able to capture and generalise metaphorical mechanisms. In th
 is paper we present the first metaphor identification method that uses rep
 resentations constructed from property norms. Such norms have been previou
 sly shown to provide a cognitively plausible representation of concepts in
  terms of semantic properties. Our results demonstrate that such property-
 based semantic representations provide a superior model of cross-domain kn
 owledge projection in metaphors\, outperforming standard distributional mo
 dels on a metaphor identification task. \n\n=============\n\nSpeaker: Guo 
 Yu (Rainbow Graphics & Interaction group)\n\nTitle: Effects of Timing on A
 gency during Mixed-Initiative Interaction\n\nAbstract: Machine learning ha
 s made human-computer interfaces more intelligent: They can observe and in
 fer users' behaviours\, and even take some initiatives during the interact
 ion. While various applications have been developed featured with mixed-in
 itiative interaction\, few studies have looked into what factors would aff
 ect users' sense of control and how. In this talk\, I will introduce our s
 tudy on the effects of timing on users' agency perception. We hypothesised
  that rhythmic intervals during the interaction could positively affect us
 ers' perceived agency\, entrainment behaviours\, performance and relaxatio
 n\, while arrhythmic could be damaging on all those aspects. We designed a
 nd carried out two within-subjects experiments\, one used visual stimuli a
 nd the other used auditory stimuli. The results have supported our hypothe
 ses. This study provided quantitative measures of timing pattern during ba
 ck-and-forth interaction\, and the resulting insights could be used to inf
 orm the design of mixed-initiative systems\, such as programming-by-exampl
 e and probabilistic programming of end-user automation.\n
LOCATION:Computer Laboratory\, William Gates Building\, Room FW26
END:VEVENT
END:VCALENDAR