Related Publications

Thank you for visiting our publications page.  This page will be updated with our project publications throughout the duration of the Slándáil project.

Ontology, Conceptual Systems and Terminology – A Disaster Case Study

Authors:

Xiubo Zhang, Khurshid Ahmad

Publication:

In DISONTOLOGY 2015 workshop. In the Book of Abstracts of the 20th European Symposium on Languages for Special Purposes, 8-10 July 2015, Vienna, Austria

Abstract:

Thorough understanding of the knowledge structures involved in emergency situations is crucial to the effective and efficient dissemination of public information during disaster events. Such understanding has traditionally been encoded manually by knowledge engineers in the form of ontologies from the insights supplied by domain experts. Despite the superior quality of the ontologies produced, however, the high costs associated with the human labours in this method limits its applicability on a larger scale. In this work, we present an alternative computerassisted workflow which is capable of producing terminologies and ontologies for the disaster management domain utilising lexical information extracted from large quantities of disaster-themed texts gathered from both authoritative and social media sources. The proposed workflow has been put to test in the Slandail project where algorithms first extract from a corpus comprising messages disseminated to the public by agencies such as FEMA (US), EMA (Australian), NCDEM (New Zealand) and Public Safety Canada (Canada) candidate terminologies and ontologies for disaster management, which are then revised and enhanced by knowledge engineers. The resulting terminology and ontology is presented as in the Slandail Terminology Wiki.

Download the DISONTOLOGY book of abstracts

Disaster Terminology Creation and Maintenance: A multi-lingual case study

Authors:

Panizzon, Raffaella, Zhao, Zeyan

Publication:

In DISONTOLOGY 2015 workshop. In the Book of Abstracts of the 20th European Symposium on Languages for Special Purposes, 8-10 July 2015, Vienna, Austria

Abstract:

When disaster strikes, emergency management operators are compelled to respond with timely and adequate measures. In this scenario, the management of incoming and outgoing information as well as communication from/to citizens and within agencies is paramount. Our EU project Slándáil (Security System for Language and Image Analysis) attempts to support emergency management agencies and the civil society at large by providing an automated text and image analytics system that can be used before, during and after disasters to enhance dissemination of public and inter-agency information. To this end, a trilingual (English, Italian and German) disaster lexicon was created by extracting terminology from corpora of texts – such as agencies’ technical reports, public information material, news outlets and social media – identified as central to natural disasters, disaster relief, disaster recovery and disaster resilience.

Both term extraction and knowledge representation have been informed by the need to cater simultaneously for the interests and requirements of multiple audiences ranging from emergency management agencies through voluntary organisations to lay people. To this end, a number of publicly available termbases and termwikis have been explored with a view to identifying best practices.

All the terms collected are now displayed on the Slándáil Terminology Wiki. Each entry is designed according to international principles of terminology management, with special reference to ISO Standards. Accordingly, our wiki provides main terms, their synonyms, variants, abbreviations and acronyms as well as linguistic, conceptual and encyclopaedic information. In this paper we investigate issues of term extraction and knowledge representation in termwikis as well as initial responses and challenges to our work so far.

Download the DISONTOLOGY book of abstracts

Propagating Disaster Warnings on Social and Digital Media

Authors:

Stephen Kelly, Khurshid Ahmad

Publication:

In Proceeding of Intelligent Data Engineering and Automated Learning – IDEAL 2015. Volume 9375 of the series Lecture Notes in Computer Science pp 475-484

Abstract:

A nexus of techniques including information extraction techniques, including a bag of words model, web and social media search and time series analysis, are discussed that may reveal the potential of social media and social networks. Social aspects of data privacy are discussed to ensuring that the data collected, filtered, and then used. This work is the effort of Trinity College Dublin and other universities.

Purchase publication here.

Ethical Framework For a Disaster Management Decision Support System which Harvests Social Media Data on a Large Scale

Authors:

Damian Jackson, Carlo Aldrovandi, Paul Hayes

Publication:

In Proceeding of Information Systems for Crisis Response and Management in Mediterranean Countries. Volume 233 of the series Lecture Notes in Business Information Processing pp 167-180

Abstract:

This paper presents preliminary results of ongoing research on the ethics of using social media on a large scale in disaster management.

To date social media use by disaster response agencies has been relatively ad-hoc. The Slándáil project aims to build a system for harvesting publicly available data from social media and using it in an ethically responsible and appropriate way to enhance the response of emergency services to natural disaster.

The ethical framework draws on the traditions of Isaiah Berlin’s value pluralism and Giorgio Agamben’s State of Exception in its approach. Value pluralism relates to an understanding that every pluralist society is organized around several and different sets of values and traditions. State of Exception theory is concerned with ethical consequences that arise when governments or state agencies arrogate to themselves extra powers in response to extraordinary circumstances, such as a natural disaster.

The implications of these ethical approaches for the Slándáil system are examined and discussed according to their impact on the various stakeholders: the system end-users, the public at large, the state and the emergency responders themselves. Implications for the technical design and governance of the system are also deduced and evaluated.

Purchase publication here.

Modes of Communication in Social Media for Emergency Management

Authors:

Sabine Gründer-Fahrer and Antje Schlaf

Publication:

In Proceeding of the 2nd Workshop on Natural Language Processing for Computer-Mediated Communication / Social Media at the International Conference of the German Society for Computer Linguistics and Language Technology in Essen.

Abstract:

The paper examines how social media were used during the flood 2013 in Central Europe and what differences in use appeared among different kinds of media. We found that Twitter played its most important part in exchange of current and factual information on the state of the event while Facebook prevalently was used for emotional support and organization of volunteers help. In a corpus-based comparative study we show how the different communicative modes prevalent in the registers German Facebook, Twitter and News are clearly reflected by the characteristic content, conceptualization and language of the respective register. The methods used include differential analysis, sentiment analysis, topic modeling, latent semantic analysis and distance matrix comparison.

Download publication here.

Ontology and the Illusion of Knowledge: Mines of text and nuggets of enlightenment

Authors:

Khurshid Ahmad

Publication:

In Proceeding of Terminology and Knowledge Engineering 2014, At Berlin, Germany

Abstract:

Ontology and the Illusion of Knowledge: Mines of text and nuggets of enlightenment
Ontologies are proliferating as life – we have gene-, plant-, disease-, pharmaceuticals are being developed essentially (a) to standardise the terminology of genes, plants, disease and pharmaceuticals and (b) to create networks of relationships between terms within and across disciplines for reasoning tasks.

Work and leisure ontologies are not far behind: the widespread use of (inter-related) financial instruments, sometimes with disastrous consequences for the bulk of humanity, has encouraged the finance community to look at the nomenclature and relationship between the instruments, business entities and processes, and legal regulations. Sporting events have become quite complex and the interest in these events transcends national and class boundaries leading to the standardisation of terms and relationships for named entities and events.

The work in ontology of art is one where a question related to classical ontology and philosophy has surfaced again, and not quite addressed in the above mentioned ontological investigation into living beings, work and leisure: Do all works of art belong to one basic ontological category? Most of the above investigations involve subject experts, information scientists, formalists/logicians and computer scientists: The description of the ‘basic’ category is often idiosyncratic of those constructing the ontology.

Those of us who have argued that the use of domain-specific corpora of diachronically will provide more objective evidence of how terms originate, and subsequently used, and how the relationship between terms emerges, have only succeeded in providing a snapshot of the ontology of a domain.
In this talk I will describe the state of text-based study of ontology and how this hermeneutical approach to building smart information is faring.

Download publication here.

BIOLOGICALLY MOTIVATED SPIRAL ARCHITECTURE FOR FAST VIDEO PROCESSING

Authors:

Min Jing, Sonya Coleman, Bryan Scotney

Publication:

In Proceeding of The International Conference on Image Processing (ICIP), Quebec City, Canada. IEEE. 5 pp.

Abstract:

Fast image processing is a key element in achieving real-time image and video analysis. The spiral addressing scheme has been an efficient tool for hexagonal image processing (HIP), whereby the image pixel indices are stored in a one- dimensional vector that enables fast processing. Unlike HIP, which requires a complex resampling scheme, we present a novel “squiral” (square spiral) image processing (SIP) frame- work that provides a spiral addressing scheme for direct ap- plication to standard square pixel-based images. A SIP-based non-overlapping convolution technique is developed by simu- lating the eye tremor phenomenon of the human visual system to accelerate computation in feature extraction. Furthermore, we deploy the proposed simulated eye tremor technique on a sequence of video frames. The preliminary results based on two action video clips demonstrate the potential of the SIP- based eye tremor model to facilitate fast video processing.

Download publication here.

Trust-Building through Social Media Communications in Disaster Management

Authors:

Maria Grazia Busà, Maria Teresa Musacchio, Shane Finan, Cilian Fennel

Publication:

In Proceeding of SWDM’15

Abstract:

Social media provides a digital space – a meeting place, for different people, often representing one or more groups in a society. The use of this space during a disaster, especially where information needs are high and the availability of factually accurate and ethically sourced data is scarce, has increased substantially over the last 5-10 years. This paper attempts to address communication in social media and trust between the public and figures of authority during a natural disaster in order to suggest communication strategies that can enhance or reinforce trust between these bodies before, during and after a natural disaster.

Download publication here.

Recognition by Enhanced Bag of Words Model via Topographic ICA

Authors:

Min Jing, Hui Wang, Kathy Clawson, Sonya Coleman, Shuwei Chen, Jun Liu and Bryan Scotney. 

Publication:

In Proceeding of UCAmI 2014, LNCS 8867, p.523.

Abstract:

The Bag-of-Words (BoW) model has been increasingly applied in the field of computer vision, in which the local features are first mapped to a codebook produced by clustering method and then represented by histogram of the words. One of drawbacks in BoW model is that the orderless histogram ignores the valuable spatial relationships among the features. In this study, we propose a novel framework based on a topographic independent component analysis (TICA), which enables the geometrically nearby feature components to be grouped together thereby bridge the semantic gap in BoW model. In addition, the compact feature obtained from TICA helps to build an efficient codebook. Furthermore, we introduce a new closeness measurement based on Neighbourhood Counting Measure (NCM) to improve the k Nearest Neighbour classification. The preliminary results based on KTH and Trecvid data demonstrate the proposed TICA/NCM approach increases the recognition accuracy and improve the efficiency of BoW model.

Purchase publication here.

A Novel Spiral Addressing Scheme for Rectangular Images

Authors:

Min Jing ; Univ. of Ulster, UK ; Scotney, B. ; Coleman, S. ; McGinnity, M.

Publication:

In Proceeding of 14th IAPR International Conference on Machine Vision Applications (MVA), 2015

Abstract:

Spiral architectures have been employed as an efficient addressing scheme in hexagonal image processing (HIP), whereby the image pixel indices can be stored in a one-dimensional vector that enables fast image processing. However, this computational advance of HIP is hindered by the additional time and effort required for conversion of image data to a HIP environment, as existing hardware for image capture and display are based predominantly on traditional rectangular pixels. In this paper, we present a novel spiral image processing framework that develops an efficient spiral addressing scheme for standard square images. We refer to this new framework as “squiral” (square spiral) image processing (SIP). Unlike HIP, conversion to the SIP addressing scheme can be achieved easily using an existing lattice with a Cartesian coordinate system; there is also no need to design special hexagonal image processing operators. Furthermore, we have developed a SIP-based non-overlapping convolution technique by simulating the “eye tremor” phenomenon of the human visual system, which facilitates fast computation. For illustration we have implemented this technique for the purpose of edge detection. The preliminary results demonstrate the efficiency of the SIP framework by comparison with standard 2D convolution and separable 2D convolution.

Purchase publication here.

Multiscale Squiral (Square-Spiral) Image Processing

Authors:

Min Jing, Sonya. A. Coleman, Bryan. W. Scotney, Martin McGinnity

Publication:

In Proceeding of IRISH MACHINE VISION & IMAGE PROCESSING Conference 2015

Abstract:

In this paper, we present a multiscale “squiral” (square spiral) image processing (SIP) framework. An efficient spiral addressing scheme is deployed for standard pixel based square images to facilitate fast image processing. A SIP-based convolution technique is developed by simulating the “eye tremor” phenomenon of the human visual system. The multiscale SIP operators are constructed by converting existing square image operators according to the SIP addressing scheme. The results of edge detection based on three-layer SIP images and SIP operator at four different scales demonstrate the efficiency of the proposed framework by comparison with standard 2D convolution.

Download publication here.

Terminological Variation between Flood-related News and Social Media

Authors:

Gründer-Fahrer, Sabine; Schlaf, Antje; Heyer, Gerhard

Publication:

In Book of Abstracts of the 20th European Symposium on Languages for Special Purposes, 8-10 July 2015, Vienna, Austria

Abstract:

Digitalization of media content is leading to new channels and distribution routes of information and new types of media, most prominently social media. At the same time it is converging all different and so-far disparate branches of media into one multi-modal and multi-medial digital room which consequently is going to contain an overwhelming amount of information of great diversity and makes the different types of media competing. From the point of view of emergency management the rise of social media has opened up the possibility to leave the province of official channels and get more in touch with the people.
In order to handle this flood of information in a productive way and to embark to unknown medial territories computer-based tools have become indispensable. One great challenge for Natural Language Processing (NLP) though is the fact that effectiveness of software can only be achieved by adaptation of methods to the peculiarities of different genres and domains of data as well as to different needs of certain user groups.

The paper is going to make a twofold contribution in this field, first, by giving statistically-based knowledge of the special character and new potential of social media information for emergency management; second, by promoting adaptation of NLP methods to the peculiarities of social media data.

We have statistically analyzed register variation on the level of terminology among news and Facebook documents concerned with the 2013 flood event in Germany. On the basis of log-likelihood measure we extracted terms that characterize flood-related documents in news and Facebook, respectively, and showed their co-occurrence pattern. Furthermore, we have done comparative analysis, i.e. by extracting lists of terms which represent the terminological difference between the flood documents in both registers or by measuring semantic distance between them.

Download book of abstracts here.

DIMPLE: DIsaster Management and Principled Large-scale information Extraction

This workshop addresses the use and implications of the use of technology in the management of the disaster life cycle: starting from warnings about impending disasters to suggestions about recovery. The technology issue has become more poignant with the advent of social media and its continually increasing use in disasters across the world. Consider the major disasters of this century, which has just begun, including hurricanes in the USA, earthquakes in Haiti, and tsunamis in Japan. In each of these cases there is documented evidence that social media is quite helpful in disseminating life and business critical information quickly and effectively. The citizen is playing or should play an active role in monitoring, averting and rehabilitating before, during and after a disaster. One learns from the disaster archives referenced above that there is a need for an ethically well-grounded and accessible system that can harness the limitless data that streams through the social media and the formal media.

 

Editors:

Khurshid Ahmad Trinity College Dublin, IRELAND.
Carl Vogel Trinity College, Dublin, IRELAND.

 

Workshop Organizers/Organizing Committee:

Khurshid Ahmad Trinity College Dublin, IRELAND.
Carl Vogel Trinity College, Dublin, IRELAND.

 

Workshop Programme Committee:

Khurshid Ahmad Trinity College Dublin, IRELAND.
Gerhard Heyer University of Leipzig, GERMANY
Linda Hogan Trinity College, Dublin, IRELAND.
Bodil Madsen Copenhagen Business School, DENMARK.
Sadhbh McCarthy Centre for Irish and European Security, Dublin, IRELAND.
Maria Teresa Musacchio University of Padova, ITALY.
Henrik Sørenson Copenhagen Business School, DENMARK.
Carl Vogel Trinity College, Dublin, IRELAND.

Download LREC2014 DIMPLE Workshop Proceedings.

View the presentations here.