DEGAS 2009 | Proceedings August 24th 2009 | Uppsala, Sweden Using human language technology to support the handling officers at the Swedish Social Insurance Agency Hercules Dalianis† Ola Knutsson‡ Teresa C. Pargman† †Department of Computer ‡School of Computer Science and Systems Sciences (DSV) and Communication KTH-Stockholm University KTH Forum 100, SE-164 40 Kista SE-100 44 Stockholm Sweden Sweden hercules@dsv.su.se, knutsson@csc.kth.se tessy@dsv.su.se ABSTRACT billion SEK (approximately 100 million Euro) The Swedish Social Insurance Agency, payment per day. The agency handles a large (Försäkringskassan) receives 40 000 e-mails per amount of e-mail and phone calls every day. The month as well as phone calls from the citizens handling officers have a large amount of that are handled by almost 500 handling officers. knowledge on what and how to answer citizens. To initiate the process to make their work more Nevertheless, they are in need of assistance to efficient we carried out two user-centered design cope with the great amount of e-mails and workshops with the handling officers at telephone calls they get daily; to allocate and Försäkringskassan with the objective of finding coordinate their work better and, to capitalize and in what ways human language technology might update their competence and knowledge on the facilitate their work. One of the outcomes from domain of social insurance. For this purpose, e- the workshops was that the handling officers services based on human language technology required a support tool for handling and seem to be appropriated to introduce into the answering e-mails from their customers. Three handling officers’ work, as the bulk of the e-mail main requirements were identified namely to find arriving to the Swedish Social Insurance Agency the correct template to be used in the e-mail might be answered by providing citizens with answers, a support to automatically create automatic or semi-automatic responses. IMAIL templates and finally an automatic e-mail [1] is the research project within we will study answering function. We will during two years how to introduce language technology into the focus on these design challenges within the handling officers’ work. IMAIL-project. RELATED RESEARCH Keywords During the last decade, many countries have put a Human language technology, Swedish, automatic lot of efforts in developing and introducing e- e-mail answering, user centered design services into the public sector. Sweden, for INTRODUCTION example, is one of the countries that together The Swedish Social Insurance Agency, is one of with US and Denmark shared the third position the largest paying agents in Sweden, with 1 (see Grundén, [2]). Indeed, according to the Swedish government, public organizations are considered as precursors in the introduction and propagation of e-services and information into the society. For instance, it has been suggested that public authorities should be stimulated to develop “the 24hours authority” (Statskontoret, 30 DEGAS 2009 | Proceedings August 24th 2009 | Uppsala, Sweden [3].). “The 24hours authority” is a term that see Ku et al [8]. Scheffer [9] have analyzed introduces the idea that public service and incoming e-mail to a European education information should be available to the citizens at provider, and he found that 42 percent of the any time through the use of Information incoming e-mails could be answered with nine Technology. For an overview of the different different standard answers. In Busemann et al levels of e-Government see Krogstie [4]. This [10] there is an overview on automatic e-mail particular vision of e-government puts thus high answering. demands on both employees and citizens. DEVELOPING AND DESIGNING HUMAN LANGUAGE Grundén [2], for example, showed in her surveys TECHNOLOGY FOR HANDLING OFFICERS that important implementation aspects of e- We have carried out two user centered design Government are closely interrelated with the workshops with handling officers at the Swedish competence and knowledge development of the Social Insurance Agency in Stockholm to find civil servants conducting their work with new out in what ways human language technology electronic tools. based services might support handling officers in In the same sense, Cajander & Ericsson [5] their daily tasks. The first workshop was a so- argues that e-Government will make civil called future workshop, see Löwgren & servants ill-healthy since they can not affect their Stolterman [11], and focused on the design of a working situation since they are captured in the future system for e-mail handling at the Swedish way their computer system work and they Social Insurance Agency. In the second therefore will not able to make creative solutions. workshop the handling officers created scenarios, Fully aware of these risks, our work intends to storyboards and sketches based on design themes provide both, employees and thus citizens with from the future workshop. usable and quality–based language technology One of the outcomes from the design workshops e-services. was that handling officers need a support system The vision of the IMAIL project is that human for assisting them with answering e-mails. More language technology can play an important role concretely, they need a tool helping them to find in the development of user-centred services for e- the correct templates to be used in the e-mail- government. Human language technology (HLT) answers as well as support able to automatically create templates for them. When it comes to the includes all algorithms and tools that deal with fully automatic answering of the e-mails, the human speech and writing. The technology is basically based on the analysis, filtering or handling officers like the idea of getting rid of the frequent short questions, but at the same time generation of human language, but the they are very concern with how the messages applications are far more sophisticated and useful should be designed and written. The opinions of (e.g. machine translation, predictive text, speech interfaces, search engines). HLT has a clear the officers were that the messages should clearly indicate that they are answers from a machine; potential not only to support human-machine they should include a disclaimer, and always give interaction but also to support human-human a reference for how to get personal service from a interaction. real person. What kind of questions that are There exist several examples of the usage of suitable for automatic answering was also an human language technology in e-Government important issue. including for example e-mail classification for automatic routing to appropriate official see In this sense, this position paper proposes a system based on human language technology that Segev & Gal [6], human language technology as an aid in rule-making processes in Cardie et al is expected to answer as many as 30 percent of [7], and crime information extraction based on the e-mail flow received by the handling officers. More specifically, the research questions at the language technology based extraction techniques core of our project are the following: from police reports and witness narrative reports, 31 DEGAS 2009 | Proceedings August 24th 2009 | Uppsala, Sweden i) How should such a system be designed to revisited. Proceedings of the NordicCHI'06 support the handling officers in their work in Workshop User Involvement and representation the best way? in e-Government projects, Oslo, Norway ii) How can we assist the handling officers to 5.Cajander, Å. and E. Eriksson. (2007). capitalize on (reuse) and update the answers Automation and E-government Services: A already sent? Widened Perspective. The 1st International iii) How should the introduction of a new tool be Workshop on Design & Evaluation of e- integrated into the current electronic Government Applications and Services: practices shared by handling officers? DEGAS, 2007, Rio de Janeiro iv) Which are the new tasks that such a new tool may generate in the daily handling officers’ 6.Segev, A and A. Gal (2007). Enhancing work? portability with multilingual ontology-based knowledge management. Decision Support QUESTIONS FOR DISCUSSION Systems, Elsevier, in press, available at • Will the handling officer really gain of having www.sciencedirect.com. a system that assists them in answering the questions? 7.Cardie, C., Farina, C., Rawding, M. and A. Aijaz. (2008) An eRulemaking Corpus: • How large percentage of the e-mail questions Identifying Substantive Issues in Public to the Swedish Social Insurance Agency can Comments. Proceedings of the Sixth be answered automatically? International Conference on Language • Is it possibly to automatically or at least semi- Resources and Evaluation (LREC 2008), automatically create templates for answering Marrakech, Morocco, 2008. e-mails? 8.Ku, C. H., A. Iriberri, and G. Leroy (2008). • What types of citizens’ requests are suitable to Natural Language Processing and e- get an automatic or semiautomatic answer? Government: Crime Information Extraction • How should we balance efficiency and quality from Heterogeneous Data Sources, In when developing e-services based on language Proceedings of the Ninth International technology? Conference on Digital Government Research • Is it possibly to automatically or at least semi- (DG.O 2008), May 18-21, 2008, Montreal, automatically create templates for answering Canada. e-mails? 9.Scheffer, T. (2004) Email answering assistance REFERENCES by semi-supervised text classification, 1.IMAIL project home page, Intelligent Data Analysis, v.8 n.5, p.481-493, http://syslab.dsv.su.se/group/IMAIL October 2004 2.Grundén, K. (2009). E-Governmentality: On 10.Busemann S., S. Schmeier, and R. G. Arens. Electronic Administration in Local (2000). Message classification in the call center. Government. Electronic Journal of e- Proceedings of the sixth conference on Applied Government. Volume 7 Issue 1 2009 (55-64). natural language processing, Seattle, Washington, ACL, pp. 158-165. 3.STATSKONTORET (2000). 24-timmars- myndighet. Förslag till kriterier för statlig 11.Löwgren, J. and E. Stolterman (2004). elektronisk förvaltning i medborgarnas tjänst. Thoughtful interaction design: A design Stockholm. perspective on information technology. Cambridge, Mass.: MIT Press. 4.Krogstie, J 2006. Citizens, from consumers to prosumers: e-Government Services typologies 32