Kalina's Current Work and Past Projects

I hold an EPSRC Career Acceleration Fellowship, working on personalised summarisation of social media. I am currently the PI of the EC-funded TrendMiner and DecarboNet projects at Sheffield, and a Co-I of the uComp project. Earlier in 2013 I completed leading the JISC-funded EnviLOD project. Until September 2011, I was a Co-Investigator on the JISC-funded TextVRE project and the JISC-EPSRC funded GATE Cloud Exploratory: Adapting the General Architecture for Text Engineering to Cloud Computing

Between January 2006 and March 2010 I was the Principal Investigator on 3 EU-funded projects (MUSING, TAO, and ServiceFinder) and also the co-ordinator of the TAO consortium, which involved 7 partner instritutions.

Until March 2006, I was Sheffield's technical project manager and researcher on the SEKT Integrated Project, working on semantically-enabled knowledge technologies.

In addition, I have been working on the GATE infrastructure since 1999, as a member of the GATE team under the management of Prof. Hamish Cunningham. Previously I also contributed to the AKT - Advanced Knowledge Technologies - project.

Between 2001 and 2004, I was the technical manager and researcher on the MIAKT e-science project, where I was bringing together my work on GATE, AKT, and language generation, by integrating and developing language generation tools in GATE for use in knowledge management and e-science applications.

Further details:

uComp: Embedded Human Computation for Knowledge Extraction and Evaluation

The project is building upon the emerging field of Human Computation (HC) in the tradition of games with a purpose and crowdsourcing marketplaces. It is developing a scalable and generic HC framework for knowledge extraction and evaluation, as well as collection of linguistic resources (e.g. sentiment lexicons, named entity annotated social media corpora). A core part of the HC framework is delegating the most challenging tasks to large communities of users and continuously learning from their feedback to optimise automated methods as part of an iterative process.

DecarboNet

DecarboNet is building a decarbonisation platform for translating collective awareness of climate change into behavioural change. Our role is to analyse social media content for events, opinions, and arguments around climate change, making use of relevant knowledge in Linked Open Data.

Personalised Summarisation of Social Media

An EPSRC Career Acceleration Fellowship (budget £560k). I investigate and evaluate methods for personalised, abstractive multi-document summarisation across different social media. For example, diachronic summaries that combine Twitter posts, blog articles, and Facebook wall messages on a given topic. A key novel contribution of the work will be personalising the summaries, based on a model of user interests, goals, and social context, and thus help users with sensemaking and content interpretation.

TrendMiner: Large-scale, Cross-lingual Trend Mining and Summarisation of Realtime Media Streams

An EU-funded STREP project, where I am one of the principal investigators (Sheffield budget 492k euro; project budget 2.1 million euro). We are carrying out research on information extraction and extractive, interactive summarisation methods for social media. Duration: Nov 2011 - Oct 2014.

EnviLOD: Semantic Enrichment and Semantic Search with Linked Open Data (LOD)

I lead the development of GATE-based tools for efficient LOD-based semantic enrichment and evaluated these, both quantitatively and with end-users and other stakeholders. I also co-designed and evaluated an intuitive user interface for semantic search, that hides the complexities of the SPARQL semantic search language, while allowing users to search successfully, using LOD vocabularies.

GateCloud Exploratory

An EPSRC/JISC funded project, where I was a Co-Investigator (budget £71k). Research on adapting the General Architecture for Text Engineering to Cloud Computing. Running demonstrator at http://gatecloud.net. For details see this journal paper on GateCloud. Duration: Feb 2011 - Oct 2011.

TextVRE

A JISC-funded Project with KCL CeRch, School of Humanities, Sheffield, Gottingen Libraries
Co-Investigator of the Sheffield team (budget £24k). Research on a virtual research environment for e-humanities textual studies, providing web-based automatic text annotation services and a virtual research environment to facilitate collaboration and text analysis.

MUSING

MUSING EU-funded Integrated Project on the Next Generation Business Intelligence
I am the Principal Investigator of the Sheffield grant (1,120,614.30 euros) within the project, which involves 15 other partners. The research is focused on ontology-based Information Extraction, with practical applications in financial risk management, internationalisation, and IT operational risk management. The project will run for 4 years aiming to deliver initial results in the next 18 months to align with Basel II.

GATE

GATE is a General Architecture for Text Engineering developed at the Natural Language Processing Group, Department of Computer Science, University of Sheffield . GATE is a project oriented towards designing and building robust and reusable open-source tools for natural language processing, which are part of a generic framework and visual development environment. GATE offers re-usable language, processing, and visual resources, written in Java, which can be plugged easily into any other application. For an demo of our re-usable syntax-tree visual resource, see my local demo. To see some more examples of what GATE can do, try the GATE demos page.

TAO

TAO EU-funded Strategic Targeted Research Project on Transitioning Applications to Ontologies
I am the project co-ordinator and the Principal Investigator of the Sheffield grant (839,670 euros), which involves 6 other partners. GATE is being used as a case study in this project, which aims to assist the migration of existing legacy applications to open, semantics-based service architectures. In addition, customised information extraction tools are being developed, in order to provide semantic-based access to continuously updated content, such as software documentation and code, mailing list postings, database content, and industry standards. The project results are applied successfully to mining information from patents in a commercially funded project.

ServiceFinder

ServiceFinder EU-funded Strategic Targeted Research Project on Web Service Discovery
I am the project Principal Investigator of the Sheffield grant (244,000 euros), which involves 4 other partners. Research on text mining and annotation of web services to aid their automatic discovery.

SEKT

SEKT is an EU-funded project, where we are working on language processing techniques for the Semantic Web, i.e., ontology-based information extraction and natural language generation for intelligent knowledge access. SEKT will deliver software to: semi-automatically learn ontologies and extract metadata, and to maintain and evolve the ontologies and metadata over time; to provide knowledge access; besides middleware to effect integration of all the SEKT components.

HYLITE+

HYLITE+ is a natural language generation I built as part of my PhD research. This project aimed at studying the requirements, and developed an adequate approach to, the problem of generating adaptive content of dynamic hypertext documents. The work builds upon existing Natural Language Generation (NLG) techniques which account for text coherence and presentation of multiple modalities (e.g., text, graphics). In order to make the generated hypertext sensitive to different types of users, the system employs ViewGen - a belief modelling system developed at New Mexico State University and Sheffield, which represents the beliefs (the propositions held by an agent to be true), intentions and goals of dialogue participants. The results were implemented in a proof-of-concept system with a WWW interface to the user.

Back to top...

Back Home...