Journal for Language Technology and Computational Linguistics (JLCL)

Eger, Steffen

English

Investigating lexical competition – An Empirical Case Study of the German Spelling Reform of 1996/2004/2006

Zajontz, Yvonne

Kuhn, Marc

Kollmann, Vanessa

Markteffizienz durch Translation Memory Systeme? Intelligente Übersetzungstechnologien zur Reduktion von Transaktionskosten international agierender Unternehmen

Richter, Frank

Fritzinger, Fabienne

Weller, Marion

Who Can See the Forest for the Trees? Extracting Multiword Negative Polarity Items from Dependency-Parsed Text

Santini, Marina

Rehm, Georg

Sharoff, Serge

Mehler, Alexander

Marina Santini, Georg Rehm, Serge Sharoff, Alexander Mehler
Editorial
In recent years, a multitude of most interesting research has been carried out in
linguistics, psycholinguistics, computational linguistics and information retrieval with
regard to the topic of web genres. Despite the increasing interest in this novel and
innovative field within different communities, there is still a significant lack in literature,
especially concerning edited collections and journal issues that provide an overview of
recent research. The aim of this special issue of the Journal for Language Technology
and Computational Linguistics is to contribute to filling this gap. More specifically, this
issue is dedicated to automatic genre identification.
Genres are categories that subsume texts which have multiple features in common,
most importantly, a shared communicative purpose. Genres form and evolve within
specific discourse communities, are instantiated as well as enforced and most often also
given a name by members of their respective discourse communities. An important
characteristic is that their users are able to recognise certain genres, such as, for
example, an invoice, a business letter, a shopping list, or a menu very quickly based
on genre-specific properties (such as, for example, a conventionalized text structure,
certain key words, a specific layout and several other properties). Recognizing that a
text belongs to a certain genre helps in the assessment of its importance and significance
as well as the communicative goals its original author associated with the text. In other
words, genres are a very important means for effective communication.
Karlgren and Cutting () and Kessler et al. () were among the first to suggest
that genres might be useful to enhance information retrieval systems. The automatic
identification of text genres could be used to generate additional metadata about text
collections that, in turn, could enable users to search for texts based on their genre
properties (with hypothetical queries such as “find instances of the genres invoice or
business letter that contain company car ”). Later on, when it became evident that
computational linguistics had a new, exciting and most challenging target in the form
of the World Wide Web, this idea was extended to web documents (see, for example,
Crowston and Williams, ; Haas and Grams, ; Roussinov et al., ; Rehm,
). If we could identify automatically the genres of web documents we could extend
the procedure and functionality sketched above to the whole web, giving users new and
innovative means of locating relevant content online.
There is a plethora of open questions with regard to genre and web genre research.
People who approach the field from a linguistic and text linguistic perspective try
to find ways of describing and representing genres and web genres with the help of
knowledge representation formalisms such as ontologies. Another interesting question
concerns the evolution of genres and web genres, how they are formed in dynamic social
communication system and discourse communities and how certain optional properties
It is interesting to note that most articles in this issue are authored or co-authored by students
who are still engaged in ongoing research or who have recently completed their research projects.
JLCL
Editorial
or features of genre instances slowly change their status into obligatory components.
Researchers who work in psycholinguistics are, for example, concerned with the problem
of genre recognition: how do we identify instances of certain genres, what kind of cues
or inherent properties of a document do we employ in order to categorize a specific
text into a certain genre? Is it primarily words that we use for this process or layout
features or probably a significant document structure? Do ordinary web users even
think in terms of “genres”, do they categorize web documents into different types?
Computational linguists try to fit all of these currently still fragmented pieces and
insights together in order to build systems that are able to identify genres and especially
web genres automatically. A few of these open questions are of utmost importance: how
do web genres work in the hypertext environment of the World Wide Web with regard
to the document, sub-document and super-document level? What kind of features
could or should be used and extracted from proper documents in order to compute
their respective web genres? Can scalable systems be built that are able to identify
hundreds of different genres – both traditional genres and genuine web genres? How can
a reference corpus of web genres be built so that the underlying theoretical assumptions
inherently encoded into the corpus are met by the probably differing theoretical opinions
of other researchers?
This special issue and other activities its editors and contributors are involved in has
its origin in . Intense networking among members of the genre community lead
to a number of events. More precisely, in July , Marina Santini and Serge Sharoff
organised the colloquium “Towards a Reference Corpus of Web Genres” (Santini and
Sharoff, ) which was held at Corpus Linguistics  in Birmingham, UK. Shortly
afterwards, Marina Santini and Georg Rehm organised a follow-up workshop, “Towards
Genre-Enabled Search Engines: The Impact of Natural Language Processing” (Rehm
and Santini, ) which was held in conjunction with the conference Recent Advances
in Natural Language Processing (RANLP) in Borovets, Bulgaria, in September .
One result of this workshop was a common paper on the construction of a reference
corpus of web genres (Rehm et al., ). Yet another event organised in  was
the panel arranged by Mark Rosso, “Towards the Use of Genre to Improve Search in
Digital Libraries: Where Do We Go from Here?”. This panel was sponsored by SIG-CR
and SIG-HCI and held in conjunction with ASIST  in Milwaukee, Wisconsin (USA).
It lead to the publication of the special issue “Bringing genre into focus” (Freund and
Ringlstetter, ). At roughly the same time, the Scandinavian Journal of Information
Systems published a special issue titled “Genre Lens on Information Systems” (Päivärinta
et al., ) which brings into focus an additional perspective on genre usefulness.
The interest in genre research has not declined over the years, the year  is
also rich of genre-related activities. First, the workshop “Automated Document Genre
Classification Workshop: Supporting Digital Curation, Information Retrieval, and
http://corpus.leeds.ac.uk/serge/webgenres/colloquium/
http://www.sics.se/use/genre-ws/
http://www.asis.org/Conferences/AM07/panels/18.html
http://iris.cs.aau.dk/index.php/volume-20-40200841-no-1.html
Volume 25 (1) – 2009 iii
Santini, Rehm, Sharoff, Mehler
Knowledge Extraction” continues the general discussion and tries to establish a common
roadmap for future projects. Second, the forthcoming book Genres on the Web:
Computational Models and Empirical studies (Mehler et al., Forthcoming) presents a
wide range of conceptualisations of the notion of web genre together with an overview of
computational approaches to web genre classification and structure modelling. Finally,
this special issue of the Journal for Language Technology and Computational Linguistics,
which has been conceived as the book’s companion volume, is now available online with
a collection of recent genre research. Taken together, the book and this JLCL special
issue show the most up to date and comprehensive range of theoretical, computational
and empirical research on web genres.
Structure of this Special Issue
Research on the automatic identification of web genres usually falls in one of three
different categories, i. e., theoretical approaches, implementations, and the creation of
resources for analysis and evaluation.
Theoretical approaches describe the extremely hard problem that web genres are able
both to span multiple documents (i. e., a hypertext) as well as to instantiate only a small
part of a single web page. There are, however, evolutionary processes at work that are
responsible for the existence of multiple conventions and preferred instances of genres
and web genres (see, for example, Rehm, ; Mehler, ). Due to these evolutionary
processes, web genres come into being and, afterwards, slowly and continuously change.
In their article “The Evolution of Genre in Wikipedia”, Malcolm Clark, Ian Ruthven
and Patrik O’Brian Holt examine how genres develop over a period of six years in the
online encyclopedia Wikipedia. The authors concentrate on the biographical article and
generate a number of follow-up research questions as well as plans for experimental work.
A related area of research is addressed by Philip M. McCarthy, John C. Myers, Stephen
W. Briner, Arthur C. Graesser and Danielle S. McNamara who present “A Psychological
and Computational Study of Sub-Sentential Genre Recognition”. In their article the
authors describe three experiments on genre recognition based on words alone. Their
conclusions not only have implications for research on the automatic identification of
web genres but also for a better understanding of genre recognition in general.
Implementations are concerned with the creation of actual systems that accomplish
the overall goal of this field of research. When building a genre identification system,
the decision of which features to use for this process is very important. In their paper
“Cost-Sensitive Feature Extraction and Selection in Genre Classification”, Ryan Levering
and Michal Cutler describe a complex approach for the automatic selection of features
for the task of genre identification and report experimental results on two datasets.
Chaker Jebari concentrates on processes that are able to categorize documents into
multiple genres. In his paper “A New Centroid-based Approach for Genre Categorization
of Web Pages”, Jebari uses machine learning algorithms in order to compute multiple
http://www.dcc.ac.uk/events/genre-classication-2009/
http://sirao.kgf.uni-frankfurt.de/webgenrebook/index.html
iv JLCL
Editorial
ranks for each documents. This approach reflects the fact that a single web page often
contains instances of multiple genres. It is this core problem of web genre research that
is also addressed by the article “Multi-Label Approaches to Web Genre Identification”.
In their paper, Vedrana Vidulin, Mitja Luštrek and Matjaž Gams extract multiple
features from web pages in order to test the performance of several classifiers for the
task of assigning multiple genre labels to a document.
The creation of resources is closely related to the two categories mentioned above
and deals with web genre corpora and datasets. In their article “Building a Corpus of
Italian Web Forums: Standard Encoding Issues and Linguistic Features”, Silvia Petri
and Mirko Tavosanis examine linguistic properties of postings in web discussion groups
and construct a corpus of these documents. They use an encoding and annotation
scheme that is based on the TEI guidelines. Undoubtedly, one of the most significant
gaps in current web genre research is the lack of a reference corpus of web genres that
interested parties could use to evaluate their own systems based on a shared resource
that was built specifically for evaluation purposes (see, for example, Rehm et al., ).
The article “Web Genre Benchmark Under Construction” by Marina Santini and Serge
Sharoff discusses this problem in detail and suggests a solution.
Acknowledgments
This special issue would not have been possible without the time and dedication of
the following reviewers: Eric Atwell, Malcom Clark, Luanne Freund, Mikael Gunnars-
son, Theresa Heyd, Marie-Paule Jacques, Yunhyong Kim, Nedim Lipka, Mark Rosso
and Efstathios Stamatatos. Further, we gratefully acknowledge support of the Ger-
man Federal Ministry of Education (BMBF) through the research project Linguistic
Networks and of the German Research Foundation (DFG) through the Excellence
Cluster  Cognitive Interaction Technology via the Project Knowledge Enhanced
Embodied Cognitive Interaction Technologies (KnowCIT), and the Research Group 
Text Technological Information Modeling via the Project A Induction of Document
Grammars for Webgenre Representation.
References
Crowston, Kevin and Williams, Marie (): “Reproduced and Emergent Genres of Com-
munication on the World-Wide Web”. In: Proceedings of the th Hawaii International
Conference on Systems Sciences (HICSS-). IEEE Computer Society, volume , pp. –.
Freund, Luanne and Ringlstetter, Christoph (): “Special Issue: Bringing Genre into Focus”.
Bulletin of the American Society for Information Science and Technology  ().
Haas, Stephanie W. and Grams, Erika S. (): “Readers, Authors, and Page Structure – A
Discussion of Four Questions Arising from a Content Analysis of Web Pages”. Journal of the
American Society for Information Science  (): pp. –.
Karlgren, Jussi and Cutting, Douglass (): “Recognizing Text Genres with Simple Metrics
Using Discriminant Analysis”. In: COLING  – The th International Conference on
Volume 25 (1) – 2009 v
Santini, Rehm, Sharoff, Mehler
Computational Linguistics. Association for Computational Linguistics, Kyoto, volume , pp.
–.
Kessler, Brett; Nunberg, Geoffrey and Schütze, Hinrich (): “Automatic Detection of Text
Genre”. In: Proceedings of the th Annual Meeting of the Association for Computational
Linguistics and the th Meeting of the European Chapter of the Association for Computational
Linguistics. San Francisco: Morgan Kaufmann, pp. –.
Mehler, Alexander (): “A quantitative graph model of social ontologies by example of
Wikipedia”. In: Towards an Information Theory of Complex Networks: Statistical Methods
and Applications, edited by Dehmer, Matthias; Emmert-Streib, Frank and Mehler, Alexander,
Boston/Basel: Birkhäuser.
Mehler, Alexander; Sharoff, Serge and Santini, Marina (editors) (Forthcoming): Genres on the
Web: Computational Models and Empirical Studies. Springer, Berlin/New York.
Päivärinta, T.; Shepherd, M.; Svensson, L. and Rossi, M. (): “Special Issue Genre Lens on
Information Systems”. Scandinavian Journal of Information Systems  ().
Rehm, Georg (): “Towards Automatic Web Genre Identification – A Corpus-Based Ap-
proach in the Domain of Academia by Example of the Academic’s Personal Homepage”. In:
Proceedings of the th Hawaii International Conference on System Sciences (HICSS-).
Big Island, Hawaii: IEEE Computer Society.
Rehm, Georg (): Hypertextsorten: Definition – Struktur – Klassifikation. Norderstedt:
Books on Demand. (PhD thesis in Applied and Computational Linguistics, Giessen University,
).
Rehm, Georg and Santini, Marina (editors) (): Proceedings of the International Workshop
Towards Genre-Enabled Search Engines: The Impact of Natural Language Processing,
Borovets, Bulgaria. Held in conjunction with RANLP .
Rehm, Georg; Santini, Marina; Mehler, Alexander; Braslavski, Pavel; Gleim, Rüdiger; Stubbe,
Andrea; Symonenko, Svetlana; Tavosanis, Mirko and Vidulin, Vedrana (): “Towards
a Reference Corpus of Web Genres for the Evaluation of Genre Identification Systems”.
In: Proceedings of the th Language Resources and Evaluation Conference (LREC ).
Marrakech, Morocco.
Roussinov, Dmitri; Crowston, Kevin; Nilan, Mike; Kwasnik, Barbara; Cai, Jin and Liu,
Xiaoyong (): “Genre Based Navigation on the Web”. In: Proceedings of the th Hawaii
International Conference on Systems Sciences (HICSS-). IEEE Computer Society.
Santini, Marina and Sharoff, Serge (editors) (): Proceedings of the Colloquium Towards
a Reference Corpus of Web Genres, Birmingham, UK. Held in conjunction with Corpus
Linguistics .
vi JLCL


Editorial

Eberle, Kurt

Integration von regel- und statistikbasierten Methoden in der maschinellen Übersetzung

Volk, Martin

The Automatic Translation of Film Subtitles. A Machine Translation Success Story?

Levering, Ryan

Cutler, Michal

Cost-Sensitive Feature Extraction and Selection in Genre Classification

Vidulin, Vedrana

Luštrek, Mitja

Gams, Matjaž

Multi-Label Approaches to Web Genre Identification

Seewald-Heeg, Uta

Härtinger, Heribert

Textsortenbezogene linguistische Untersuchungen zum Einsatz von Translation-Memory-Systemen an einem Korpus deutscher und spanischer Patentschriften

Not a member yet

Investigating lexical competition – An Empirical Case Study of the German Spelling Reform of 1996/2004/2006

Markteffizienz durch Translation Memory Systeme? Intelligente Übersetzungstechnologien zur Reduktion von Transaktionskosten international agierender Unternehmen

Who Can See the Forest for the Trees? Extracting Multiword Negative Polarity Items from Dependency-Parsed Text

Editorial

Integration von regel- und statistikbasierten Methoden in der maschinellen Übersetzung

The Automatic Translation of Film Subtitles. A Machine Translation Success Story?

Cost-Sensitive Feature Extraction and Selection in Genre Classification

Multi-Label Approaches to Web Genre Identification

Editorial

Textsortenbezogene linguistische Untersuchungen zum Einsatz von Translation-Memory-Systemen an einem Korpus deutscher und spanischer Patentschriften

231

253