• seminar

seminar

SEMINAR

For over a decade researchers at the University of New Hampshire have been involved in the exploration and deployment of in-vehicle user interfaces. This talk will provide a short review of our efforts on deploying in-vehicle user interfaces for police through our Project54 effort. Next, the talk will discuss a number of recent driving simulator-based studies in which eye tracking data was used to estimate the visual attention of the driver to the external world, the ability of the driver to control the vehicle, and the level of the driver’s cognitive load.

Bio:
Andrew Kun is associate professor of Electrical and Computer Engineering at the University of New Hampshire. His research focus is human-computer interaction in vehicles. In this area he is primarily interested in speech interaction, as well as the use of visual behavior and pupil diameter measures to assess and improve the design of user interfaces.

Date: 2015-10-16 10:30 - 12:00

Location: Room 112, Dicksonsgatan 4

Permalink

SEMINAR

John J. Camilleri (Department of Computer Science and Engineering) will defend his licentiate thesis Analysing normative contracts. On the semantic gap between natural and formal languages.

Abstract:
Normative contracts are documents written in natural language, such as English or Swedish, which describe the permissions, obligations, and prohibitions of two or more parties over a set of actions, including descriptions of the penalties which must be payed when the main norms are violated. We encounter such texts frequently in our daily lives in the form of privacy policies, software licenses, and service agreements. The length and dense linguistic style of such contracts often makes them difficult to follow for non-experts, and many people agree to these legally-binding documents without even reading them. By investigating the processing of normative texts, how they can be modelled formally using a suitable logic, and what kinds of properties can be automatically tested on our models, we hope to produce end-user tools which can take a natural language contract as input, highlight any potentially problematic clauses, and allow a user to easily ask questions about the implications of the contract, getting a meaningful answer in natural language within a reasonable amount of time. This thesis includes four research articles by the author which investigate the various components that a system such as this would require; from entity recognition and modality extraction on natural language texts, to controlled natural languages and visual diagrams as modelling interfaces, to logical formalisms which can be used for contract representation, to the different kinds of analysis possible and how this can be linked to user questions in natural language.

Opponent:
Dr. Adam Wyner, Department of Computing Science, University of Aberdeen, Scotland

Date: 2015-10-29 14:00 - 16:00

Location: Room EA, EDIT Building, Hörsalsvägen 11, Chalmers Johanneberg

Permalink

SEMINAR

The breathing lab recently installed at Stockholm University is the only existing laboratory especially designed to understand the complex and hitherto very little understood interplay of breathing and speech production and perception in human communication. The investigations carried out in this lab are located at the interface of research on multimodal interaction and speech gesture, discourse prosody and experimental phonetics. In this talk, I will describe the setup of the breathing lab and the multiparty conversational corpus we are collecting there. In addition, I will give a couple of examples of investigations we have carried out and plan to do within the Swedish Research Council project (2014-1072) Breathing in conversation.

Date: 2015-10-15 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

Det är viktigt att hitta elever med läs- och skrivsvårigheter så tidigt som möjligt i skolan. Idag upptäcks de flesta först i sena mellanstadiet eller högstadiet, vilket är alldeles för sent. En viktig orsak till att inte fler upptäcks tidigare är bristen på enkla men effektiva metoder för att bedöma läsförmågan. Genom att kombinera ögonrörelsemätning med maskininlärning har vi utvecklat en ny teknik för tidig upptäckt av läs- och skrivsvårigheter. I detta projekt kommer vi utveckla metoden tillsammans med Järfälla och Trosa kommuns så att den blir användbar i skolan.

Link to VINNOVA project:
http://www.vinnova.se/sv/Resultat/Projekt/Effekta/2013-01779/Ny-teknik-for-tidig-upptackt-av-las--och-skrivsvarigheter/

Other related links:
http://www.meetingsinternational.se/articles.php?id=224#.Ve7vcxTtmko
http://www.svd.se/ogonen-kan-ge-besked-om-dyslexi

Date: 2015-11-19 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

What does it take to bootstrap a language? How is it maintained? How does meaning emerge and become shared? I will review some of the representative literature on the so called Symbol Grounding problem and how it relates to the field of Computational Semantics. I will discuss two prominent approaches to semantics - logical and distributional - and what might be seen as their respective strengths and shortcomings. I will suggest that what most existing approaches lack is sufficient regard for the basic idea that symbols are used to do stuff in the world in interaction with others; and consequently, that their meaning is constrained by (1) their utility in abstract reasoning/planning in a particular task; and (2) semantic coordination pressures in interaction with specific interlocutors.

I then go on to sketch a model of how a structured, probabilistic ontology of object kinds and their properties - represented using the Type Theory with Records framework- can be bootstrapped from dialogue in the context of a simple collaborative referring game. The basic premise in this model is that one can learn the meaning of semantic representations by deploying them in a particular environment and observing their effect - perlocution - to learn what they mean. The model will use a combination of the DS-TTR grammar framework (Dynamic Syntax and Type Theory with Records) for dialogue processing, and Reinforcement Learning for optimisation of what to say. The hope is that one can use such a framework in simulation to test specific hypotheses about what basic mechanisms - e.g. incrementally, repair, rejection, agreement, alignment, etc. - are required for the emergence of shared meaning; but also that it can be used as a basis for developing conversational robots that can learn new concepts and adapt existing ones in interaction with humans.

Short bio:
Arash Eshghi completed his PhD "Uncommon ground: The distribution of dialogue contexts" at Queen Mary University of London in 2009. His research focuses on dialogue and dialogue modelling from a wide range of perspectives including computational methods and experiments on human-human interaction. He is currently working at Heriot Watt University's Interaction Lab on the EPSRC funded BABBLE project: domain-general methods for learning natural spoken dialogue systems, in which speech systems can be trained to interact naturally with humans, much like a child who experiments with new combinations of words to discover their usefulness.

Date: 2015-11-05 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

Multilinguality is a key feature of today’s Web, and it is this feature that we leverage and exploit in our research work at the Sapienza University of Rome’s Linguistic Computing Laboratory, which I am going to overview and showcase in this talk.

I will start by presenting BabelNet 3.5, available at http://babelnet.org, a very large multilingual encyclopedic dictionary, semantic network and knowledge base, which covers 272 languages and provides both lexicographic and encyclopedic knowledge for all the open-class parts of speech, thanks to the seamless integration of WordNet, Wikipedia, Wiktionary, OmegaWiki, Wikidata, the Open Multilingual WordNet, GeoNames and other resources.

Next, I will present Babelfy, available at http://babelfy.org, a unified approach that leverages BabelNet to jointly perform word sense disambiguation and entity linking in arbitrary languages, with performance on both tasks on a par with, or surpassing, those of task-specific state-of-the-art supervised systems.

I will also describe the Wikipedia Bitaxonomy, available at http://wibitaxonomy.org, a new approach to the construction of two taxonomies for Wikipedia, that is, the largest and most accurate currently available taxonomy of Wikipedia pages and taxonomy of categories, aligned to each other.

Finally, I will introduce BabelTag, a new video game with a purpose for mobile phones made up of several mini-games, each of which contains a gamified annotation task (e.g. determine if the image depicts a concept; are two terms semantically related? etc.). The aim is, on the one hand, to engage players and, on the other, to validate and enrich BabelNet with accurate knowledge items.

This is joint work with many, many people: Francesco Cecconi, José Camacho Collados, Tiziano Flati, Andrea Moro, Tommaso Pasini, Simone Ponzetto, Alessandro Raganato, Daniele Vannella.

Bio:
Roberto Navigli is an Associate Professor in the Department of Computer Science of the Sapienza University of Rome. He was awarded the Marco Cadoli 2007 AIIA Prize for the best doctoral thesis in Artificial Intelligence and the Marco Somalvico 2013 AIIA Prize for the best young researcher in AI. He is the first Italian recipient of an ERC Starting Grant in computer science and informatics on multilingual word sense disambiguation (2011-2016), a co-PI of a Google Focused Research Award on Natural Language Understanding and a partner of the LIDER EU project.

His research lies in the field of Natural Language Processing (including multilingual word sense disambiguation and induction, multilingual entity linking, large-scale knowledge acquisition, ontology learning from scratch, open information extraction and relation extraction). He has served as an area chair of ACL, WWW, and *SEM, and a senior program committee member of IJCAI. Currently he is an Associate Editor of the Artificial Intelligence Journal, a member of the editorial board of the Journal of Natural Language Engineering, a guest editor of the Journal of Web Semantics, and a former editorial board member of Computational Linguistics.

Date: 2015-12-17 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

Natural Logic is a deductive inference system that works directly on syntactic representations of natural language, with no intermediate translation to logical formulas. It thus side-steps problems involved in the translation of natural language to first-order logic, while being more precise than shallow inference techniques. In particular, Natural Logic focuses on monotonicity inference patterns, such as inferring "Nina has a dog" from "Nina has a bulldog" (i.e. replacing a specific term by a more general one in a positive context), or "Nina didn't get a rose" from "Nina didn't get a flower" (i.e. replacing a general term by a more specific one in a negative context).

In this talk I will present work in progress on implementing a multilingual Natural Logic engine that builds on Grammatical Framework and imports specificity hierarchies from an underlying ontology. The goal is to explore the applicability of Natural Logic to different inference patterns, and to investigate how language-independent those patterns are.

Short bio:
Christina Unger is a postdoctoral researcher in the Semantic Computing group affiliated to the Cluster of Excellence on Cognitive Interaction Technology (CITEC) at Bielefeld University. Her major research interest lies in the area of computational semantics, with a focus on ontology-based natural language understanding and question answering.

Date: 2015-12-10 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

I will describe some of our recent advances in the prediction of predicate-argument structure in natural language text.

First, I will describe a dynamic programming algorithm for efficient constrained inference in semantic role labeling. The algorithm efficiently captures a majority of the structural constraints examined by prior work in this area, which has resorted to either approximate methods or slow integer linear programming solvers. In addition, it allows for structured learning, with respect to constrained conditional likelihood, which leads to improved predictions over a locally learned model.

Second, I will describe how the potential functions in the graphical model corresponding to the dynamic program can be replaced with neural networks. In addition to increased modeling power and automatically induced feature combinations, this allows us to embed phrasal arguments and semantic roles jointly in the same vector space, and provides a flexible framework for multi-task learning by the embedding of semantic roles from multiple annotation schemes in a shared vector space.

With these advances, both by themselves and combined, we obtain state-of-the-art results on both PropBank- and FrameNet-annotated datasets.

Bio:
Oscar Täckström is a research scientist at Google in New York, where he works primarily on the semantic analysis of text and question answering from structured knowledge bases. Before joining Google in 2013, he was a PhD student in the computational linguistics group at Uppsala University and a research scientist at the Swedish Institute of Computer Science. In his thesis, he explored the use of incomplete and cross-lingual supervision for learning statistical models in natural language processing. Together with Ryan McDonald and Jakob Uszkoreit, he received the IBM Best Student Paper Award at NAACL 2012.

Date: 2015-11-26 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

Dictionaries are important tools for language learners and translators, as well as valuable cultural artefacts. Moreover, lexical knowledge is essential to high-quality systems for a variety of natural language processing tasks. One problem, however, is that language changes. New words are coined every day, and new meanings of established words commonly emerge. Dictionaries therefore need to be constantly updated, but doing so manually is very expensive. Techniques for automatically keeping them up to date are therefore required. In this talk I will present research on two topics related to this theme: 1. applying a word-sense induction system to automatically find new meanings of words, and 2. using GPS-tagged social media posts to identify previously-undocumented regionalisms.

Date: 2015-11-12 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

SEMINAR

Nearly two years have passed since I presented my work on Pengines (Prolog engines) [4] and SWISH [1,6], a web front-end for Prolog that utilises Pengines. Due to an unfortunate "burn-out" and a long sick leave I have not been able to do much work on them for the past year or so, but I do have at least a couple of months worth of work to report on, and even during my leave things have matured a lot. In particular, other people, mostly Jan Wielemaker in Amsterdam, have continued to refine the Pengines library as well as the SWISH application.

SWISH is now accessible from the SWI-Prolog web site, where it can be used to run small Prolog programs for demonstration, experimentation and education.[1,6] SWISH has also been connected with the ClioPatria semantic web toolkit, where it allows for collaborative development of programs and queries related to a dataset as well as performing maintenance tasks on the running server,[2,5] and SWISH has been embedded in the Learn Prolog Now! online Prolog book.[3,6]

Jan Wielemaker has done a tremendous job implementing these applications. Still I believe that Pengines can do a lot more than serving as the machinery underlying SWISH and similar applications. In this talk I'm going to sketch a programming language - an extension of Prolog - that I will refer to as *Web Prolog*. Borrowing features from the Erlang programming language, Web Prolog allows a neater implementation of Pengines, makes concurrent and distributed programming easier, and can possibly be standardised by the W3C.

REFERENCES:
[1] http://swish.swi-prolog.org/
[2] http://linkedpolitics.d2s.labs.vu.nl/swish
[3] http://lpn.swi-prolog.org
[4] Torbjörn Lager and Jan Wielemaker (2014) Pengines: Web Logic Programming Made Easy, In: Theory and Practice of Logic Programming, 14 (4-5) s. 539-552.
[5] Jan Wielemaker, Wouter Beek, Michiel Hildebrand, Jacco van Ossenbruggen (2015) ClioPatria: A SWI-Prolog Infrastructure for the Semantic Web, In: Semantic Web Journal, 2015.
[6] Jan Wielemaker, Torbjörn Lager, and Fabrizio Riguzzi (2015) SWISH: SWI-Prolog for Sharing. In Stefan Ellmauthaler and Claudia Schulz, editors, International Workshop on User-Oriented Logic Programming (IULP 2015).

Date: 2015-10-01 10:30 - 12:00

Location: L308, Lennart Torstenssonsgatan 8

Permalink

X
Loading