15.11.2012 Views

icegov2012 proceedings

icegov2012 proceedings

icegov2012 proceedings

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

how language technology has been or could be used in relation to<br />

some but not all of the processes in workflow management.<br />

In a Workflow Management System (WfMS), the triage and<br />

routing of cases/tasks are an important part, and automatic means<br />

could be used to make this step more efficient. If the case or the<br />

carrier of the case includes accessible text, language technology<br />

could improve this process, by routing the task to a handling<br />

officer who knows the area of the task, this could be done using<br />

text categorization. An experiment with document routing was<br />

carried out by [10]. They used a very limited part of the textual<br />

resources, such as references and subject lines, in their<br />

experiment.<br />

When looking at businesses outside government, “trouble tickets”<br />

have been analyzed [11]. One could see the use of the same<br />

techniques within a government agency, here the “trouble tickets”<br />

could for instance concern reports from the citizens about<br />

“problems” in their neighborhood, which then could be mined<br />

about problem areas of a specific kind in the whole community. In<br />

the paper by Ku et al [8] information extraction was used on<br />

police reports, newspaper articles, and victims’ and witnesses’<br />

reports. This is an interesting example of how a system could<br />

combine extracting information from sources on the “inside”<br />

(police reports, victims’ and witnesses’ reports) with the outside<br />

world (newspaper articles), and how the extraction of people,<br />

places, vehicle names and other entities from both sides could<br />

give a broader picture of an event.<br />

E-mail is a common channel to contact an organization and to<br />

initiate a case. Knutsson et al. [7] discusses the benefits of<br />

automated e-mail answering in large organizations, as well as a<br />

number of tools and techniques useful for e-mail processing in<br />

eGovernment settings. Most early e-mail answering systems used<br />

machine learning and pursued the task of text classification.<br />

Similar to the text classification task is matching incoming e-mail<br />

messages to a list of frequently asked questions (FAQs). Iwai et<br />

al. [5] makes an observation that 30-40% of an e-mail flow<br />

addresses often reoccurring topics, and presents a system for<br />

developing an FAQ list and matching these FAQs to future emails.<br />

Question-answering systems answer single-sentence questions. In<br />

closed domains dealing with one or few subjects, automated FAQ<br />

retrieval has proved practical. Recently, template based<br />

techniques have gained more attention. Samarakoon et al. [9]<br />

presents a help-desk application where each FAQ is associated<br />

with hand-written text patterns that match future queries.<br />

Statistical n-gram matching and ranking algorithms are applied. A<br />

more “intelligent” alternative to FAQ retrieval are chatbots alias<br />

conversational agents alias dialogue systems; these systems<br />

maintain a dialog with the user. Deryugina [3] lists a number of<br />

chatbot applications in customer care, which is relevant for<br />

eGovernment.<br />

3. CONCLUSIONS<br />

In this paper we have discussed the use of language technology in<br />

several different areas of workflow management. Language<br />

technology could increase efficiency by task routing and<br />

automatic answering, and some of the techniques even prevent<br />

some of the citizens’ cases from becoming part of the workflow.<br />

The use of language technology in business intelligence might<br />

increase quality by giving organizations a better picture of the<br />

496<br />

information on the inside, but also what is going on outside the<br />

government agency.<br />

4. REFERENCES<br />

[1] Abecker, A., Bernardi, A., Maus, H., Sintek, M. and Wenzel,<br />

C. Information supply for business processes: coupling<br />

workflow with document analysis and information retrieval.<br />

Knowledge-Based Systems, 13 (2000), 271-284.<br />

[2] Bae, H., Hu, W., Yoo, W. S., Kwak, B. K., Kim, Y., and<br />

Park, Y. Document configuration control processes captured<br />

in a workflow. Comput. Ind. 53, 2 (February 2004), 117-131.<br />

[3] Deryugina, O.V. Chatterbots. Scientific and Technical<br />

Information Processing, 37, 2 (2010), 143–147.<br />

[4] Esposito, F., Ferilli, S., Teresa M.A. Basile, T.M.A., and Di<br />

Mauro, N. (2008). Machine Learning for Digital Document<br />

Processing: from Layout Analysis to Metadata Extraction,<br />

Studies in Computational Intelligence (SCI) 90 (2008), 105–<br />

138.<br />

[5] Iwai, K., Iida, K., Akiyoshi, M., and Komoda, N. (2010). A<br />

help desk support system with filtering and reusing e-mails,"<br />

In Proceedings of the 8th IEEE International Conference on<br />

Industrial Informatics (INDIN 2010),(Osaka, Japan, 13-16<br />

July, 2010), 321-325.<br />

[6] Jurafsky, D. and Martin, J. H. Speech and Language<br />

Processing: An Introduction to Natural Language<br />

Processing, Computational Linguistics, and Speech<br />

Recognition, Prentice Hall, Upper Saddle River, NJ, 2008.<br />

[7] Knutsson, O., Pargman, T. C., Dalianis, H., Rosell, M., and<br />

Sneiders, E. Increasing the Efficiency and Quality of E-mail<br />

Communication in eGovernment Using Language<br />

Technology. In Proceedings of the IFIP e-government<br />

conference (EGOV 2010), (Lausanne, Switzerland, August<br />

29-September 2, 2010). Trauner Verlag, Linz, Austria, 45-<br />

52.<br />

[8] Ku, C. H., Iriberri, A., and Leroy, G. Natural Language<br />

Processing and e-Government: Crime Information Extraction<br />

from Heterogeneous Data Sources. In Proceedings of the 9th<br />

Annual International Digital Government Research<br />

Conference, (Montreal, Canada, 2008).<br />

[9] Samarakoon, L.; Kumarawadu, S.; Pulasinghe, K. Automated<br />

question answering for customer helpdesk applications. In<br />

Proceedings of 6th IEEE International Conference on<br />

Industrial and Information Systems (ICIIS 2011), (Kandy, Sri<br />

Lanka, 16-19 August 2011) IEEE, 328-333,<br />

[10] Soares, C. and Calejo, M. Intelligent Document Routing as a<br />

First Step towards Workflow Automation: A Case Study<br />

Implemented in SQL. ISoLA (1) 2010: pp. 276-284.<br />

[11] Temprado, Y., Molinero, F. J., Garcia, C. and Gomez, J.<br />

Knowledge Discovery from Trouble Ticketing Reports in a<br />

Large Telecommunication Company. In Proceedings of the<br />

2008 International Conference on Computational<br />

Intelligence for Modelling Control & Automation (CIMCA<br />

'08). IEEE Computer Society, Washington, DC, USA, 37-4

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!