Kalina Bontcheva

GATE: an Architecture for Development of Robust HLT applications

In this paper we present GATE, a framework and graphical development environment which enables users to develop and deploy language engineering components and resources in a robust fashion. The GATE architecture has enabled us not only to develop a number of successful applications for various language processing tasks (such as Information Extraction), but also to build and annotate corpora and carry out evaluations on the applications generated. The framework can be used to develop applications and resources in multiple languages, based on its thorough Unicode support.

Evolving GATE to meet new challenges in language engineering

In this paper we present recent work on GATE, a widely-used framework and graphical development environment for creating and deploying Language Engineering components and resources in a robust fashion. The GATE architecture has facilitated the development of a number of successful applications for various language processing tasks (such as Information Extraction, dialogue and summarisation), the building and annotation of corpora and the quantitative evaluations of LE applications. The focus of this paper is on recent developments in response to new challenges in Language Engineering: Semantic Web, integration with Information Retrieval and data mining, and the need for machine learning support.

Analysis of named entity recognition and linking for tweets

Applying natural language processing for mining and intelligent information access to tweets (a form of microblog) is a challenging, emerging research area. Unlike carefully authored news text and other longer content, tweets pose a number of new challenges, due to their short, noisy, context-dependent, and dynamic nature. Information extraction from tweets is typically performed in a pipeline, comprising consecutive stages of language identification, tokenisation, part-of-speech tagging, named entity recognition and entity disambiguation (e.g. with respect to DBpedia). In this work, we describe a new Twitter entity disambiguation dataset, and conduct an empirical analysis of named entity recognition and disambiguation, investigating how robust a number of state-of-the-art systems are on such noisy texts, what the main sources of error are, and which problems should be further investigated to improve the state of the art.

A natural language query interface to structured information

Accessing structured data such as that encoded in ontologies and knowledge bases can be done using either syntactically complex formal query languages like SPARQL or complicated form interfaces that require expensive customisation to each particular application domain. This paper presents the QuestIO system - a natural language interface for accessing structured information, that is domain independent and easy to use without training. It aims to bring the simplicity of Googles search interface to conceptual retrieval by automatically converting short conceptual queries into formal ones, which can then be executed against any semantic repository. QuestIO was developed specifically to be robustwith regard to language ambiguities, incomplete or syntactically ill-formed queries, by harnessing the structure of ontologies, fuzzy stringmatching, and ontology-motivated similarity metrics.

Ontology-based information extraction for business intelligence

Business Intelligence (BI) requires the acquisition and aggregation of key pieces of knowledge from multiple sources in order to provide valuable information to customers or feed statistical BI models and tools. The massive amount of information available to business analysts makes information extraction and other natural language processing tools key enablers for the acquisition and use of that semantic information. We describe the application of ontology-based extraction and merging in the context of a practical e-business application for the EU MUSING Project where the goal is to gather international company intelligence and country/region information. The results of our experiments so far are very promising and we are now in the process of building a complete end-to-end solution.

Architectural elements of language engineering robustness

We discuss robustness in LE systems from the perspective of engineering, and the predictability of both outputs and construction process that this entails. We present an architectural system that contributes to engineering robustness and low-overhead systems development (GATE, a General Architecture for Text Engineering). To verify our ideas we present results from the development of a multi-purpose cross-genre Named Entity recognition system. This system aims be robust across diverse input types, and to reduce the need for costly and timeconsuming adaptation of systems to new applications, with its capability to process texts from widely differing domains and genres.

CLOnE: controlled language for ontology editing

This paper presents a controlled language for ontology editing and a software implementation, based partly on standard NLP tools, for processing that language and manipulating an ontology. The input sentences are analysed deterministically and compositionally with respect to a given ontology, which the software consults in order to interpret the inputs semantics; this allows the user to learn fewer syntactic structures since some of them can be used to refer to either classes or instances, for example. A repeated-measures, task-based evaluation has been carried out in comparison with a well-known ontology editor; our software received favourable results for basic tasks. The paper also discusses work in progress and future plans for developing this language and tool.

Making Sense of Social Media Streams through Semantics: a Survey

Using semantic technologies for mining and intelligent information access to social media is a challenging, emerging research area. Traditional search methods are no longer able to address the more complex information seeking behaviour in media streams, which has evolved towards sense making, learning, investigation, and social search. Unlike carefully authored news text and longer web context, social media streams pose a number of new challenges, due to their large-scale, short, noisy, contextdependent, and dynamic nature. This paper defines five key research questions in this new application area, examined through a survey of state-of-the-art approaches to mining semantics from social media streams; user, network, and behaviour modelling; and intelligent, semanticbased information access. The survey includes key methods not just from the Semantic Web research field, but also from the related areas of natural language processing and user modelling. In conclusion, key outstanding challenges are discussed and new directions for research are proposed.

SVM based learning system for information extraction

This paper presents an SVM-based learning system for information extraction (IE). One distinctive feature of our system is the use of a variant of the SVM, the SVM with uneven margins, which is particularly helpful for small training datasets. In addition, our approach needs fewer SVM classifiers to be trained than other recent SVM-based systems. The paper also compares our approach to several state-of-the-art systems (including rule learning and statistical learning algorithms) on three IE benchmark datasets: CoNLL-2003, CMU seminars, and the software jobs corpus. The experimental results show that our system outperforms a recent SVM-based system on CoNLL-2003, achieves the highest score on eight out of 17 categories on the jobs corpus, and is second best on the remaining nine.

Automatic report generation from ontologies: The MIAKT approach

This paper presented an approach for automatic generation of reports from domain ontologies encoded in Semantic Web standards like OWL. The paper identifies the challenges that need to be addressed when generating text from RDF and OWL and demonstrates how the ontology is used during the different stages of the generation process. The main contribution is in showing how NLG tools that take Semantic Web ontologies as their input can be designed to minimises the portability effort, while offering better output than template-based ontology verbalisers.