Noisy unstructured text data is found in informal settings such as online chat, SMS, emails,message boards, newsgroups, blogs, wikis and web pages. Also, text produced by processing
spontaneous speech, printed text and handwritten text contains processing noise. Text produced
under such circumstances is typically highly noisy containing spelling errors, abbreviations, nonstandard
words, false starts, repetitions, missing punctuations, missing case information, pause
filling words such as "um" and "uh." Such text can be seen in large amounts in contact centers,
on-line chat rooms, OCRed text documents, SMS corpus etc. Documents with historical language
can also be considered noisy with respect to today’s knowledge about the language. Such text
contains important historical, religious and ancient medical knowledge that is useful. The theme
of the IJCAI 2007 Conference is "AI and its benefits to society." In keeping with this theme, this
workshop proposes to look at analytics of highly noisy text that is produced in everyday
applications in society.
The goal of the workshop is to focus on the problems encountered in analyzing noisy documents
coming from various sources. The nature of the text warrants moving beyond traditional text
analytics techniques. This workshop brings together a diverse group of researchers to present
current research and development in addressing this challenge. As a result of this workshop some
new real life noisy data sets have also become available to a wider research community.
We were fortunate to assemble a diverse group of researchers from the Natural Language
Processing, Machine Learning and Knowledge Management communities to help us in
organizing this workshop. The workshop call for papers had a very good response. We received
30 submissions spanning a diverse set of issues relevant to noisy text analytics. Each submission
was reviewed by at least three members of the program committee.
To encourage discussion, the workshop program is structured into topic-oriented oral and poster
sessions. In addition to the contributed papers, the program also contains a keynote address and a
panel discussion - on the topic of whether noisy text analytics is at all possible, and if it is then
how.
We would like to thank our organizing and program committees for their many invaluable inputs
and thoughtful reviews. We would like to thank Monojit Choudhury, Matthew Hurst, Ted
Pedersen and Sudeshna Sarkar for sharing noisy text datasets prepared by them. We would also
like to thank the others who pointed us to many relevant noisy text datasets. We thank the
International Association for Pattern Recognition for endorsing this workshop and instituting a
best student paper award. We would like to thank Raghuram Krishnapuram for chairing the
committee to decide the best student paper award. We also thank IBM Research for providing
financial support for the workshop.
Craig Knoblock, Daniel Lopresti, Shourya Roy, L. Venkata Subramaniam (Workshop Co-Chairs)
