Keyword Extraction in Information Retrieval

Harold Short

harold.short@kcl.ac.uk

King's College London

Marilyn Deegan

marilyn.deegan@kcl.ac.uk

King's College London

Laszlo Hunyadi

hunyadi@ling.arts.klte.hu

University of Debrecen

Paul Baker

p.baker@lancaster.ac.uk

Lancaster University

Dawn Archer

d.archer@lancaster.ac.uk

University of Central Lancashire

Tony McEnery

a.mcenery@lancaster.ac.uk

Lancaster University


Session chair: Harold Short.




Keyword extraction: an Overview

Laszlo Hunyadi

With the ever increasing amount of information made available in all areas of life, including economy, science, education and culture, there is an imperative need to retrieve, elaborate and present this information in the most optimal way. Since the bulk of information is textual, information retrieval is mainly concerned with texts. This talk will present an outline of the approaches, principles and techniques used in textual data retrieval based on keyword extraction. There will be an analysis of the capabilities of various approaches showing their appropriate uses.





Keyword extraction in the project Forced Migration Online

Marilyn Deegan

Forced Migration Online () provides instant access to a wide variety of online resources dealing with the situation of forced migrants worldwide. Being one of the most comprehensive textual resources dealing with humanitarian issues across a large number of countries, it faces the challenges of multilingualism, multiculturism and the essential requirement to be up-to-date and informative. That is why the organisation and retrieval of textual information as well as operability have high priority in the design and functioning of the system. This talk presents the essentials of this online resource, including its pioneering beginnings and view of future development.





Querying Keywords: Questions of difference, frequency and sense

Paul Baker

This paper examines issues to do with interpreting keyword lists  (Scott
1999), such as over-attending to lexical differences whilst ignoring
differences in word usage and/or similarities between texts. Using a
variety of technniques (e.g. analysis of key clusters or annotated
data), I show how researchers can use keyword analyses to obtain a more
accurate picture of the distinctive features of their texts or corpora.





Love - a familiar or a devil? An exploration of key domains in Shakespeare's Comedies and Tragedies

Dawn Archer, Jonathan Culpeper, Paul Rayson

Love is a common theme in Shakespeare's works. In this paper, we show
   how the UCREL Semantic Annotation Scheme (henceforth USAS), a software
program for automatic dictionary-based content analysis, can help us to
explore the semantic field of 'love' within a selection of Shakespeare's
   plays. Specifically, we will explore 3 love-tragedies (Othello, Antony
      and Cleopatra, and Romeo and Juliet) and 3 love-comedies (A Midsummer
         Night's Dream, The Two Gentlemen of Verona and As You Like It) to
determine differences in their (re)presentation of love. We will also
discuss how the semantic field of love co-occurs with different
domains in the plays, and assess the implications this has on our
understanding of love as a concept. This research builds on (i)
   Jonathan Culpeper's work on keywords in Shakespeare, using Wordsmith (Culpeper 2002), (ii) Paul Rayson's comparisons of key word and key
domain analysis  (Rayson 2003), and (iii) Dawn Archer and Paul Rayson's
work on the identification of key domains in refugee literature, using
   USAS (Archer and Rayson forthcoming).





Key words and the analysis of discourses in historical contexts

Tony McEnery

This paper examines the use of keywords to approach the discourse of
moral panic evident in the writings of the Society for the Reformation
of Manners in late seventeenth/early eighteenth century England. The
keyword approach, I will argue allows one to populate a model of moral
panic discourse, while simultaneously showing how, in that historical
context, links were forged between concepts which, while unlinked then,
have become naturalised as being linked in modern English. By showing
how keywords relate to discourse, and ultimately to a process whereby
meanings and objects become linked, the paper will argue that keywords
are important tools for the historical linguist in studying the shifting
patterns of word association in language.




Bibliography


Archer, D.
Rayson, P.
Using the UCREL automated semantic analysis system to investigate differing concerns in refugee literature
Deegan, M.
Hunyadi, L.
Short, H.
The Keyword Project: Unlocking Content Through Computational Linguistics
Office for Humanities Communication Publications
Forthcoming

Culpeper, J.
Computers, language and characterisation: An analysis of six characters in Romeo and Juliet
Melander Marttala, Ulla
Ostman, Carin
Kyto, Merja
Papers from the ASLA symposium, Conversation in life and literature
Association Suedoise de Linguistique Appliquee
Uppsala
2002
11-30

Rayson, P.
Matrix: A statistical method and software tool for linguistic analysis through corpus comparison
Ph.D. thesis, Lancaster University