Customer care in technical domains is increasingly based on e-mail communication, allowing for the reproduction of approved solutions. Identifying the customer's problem is often time-consuming, as the problem space changes if new products are launched. This paper describes a new approach to the classification of e-mail requests based on shallow text processing and machine learning techniques. It is implemented within an assistance system for call center agents that is used in a commercial setting.

Appointment scheduling is a problem faced daily by many individuals and organizations. Cooperating agent systems have been developed to partially automate this task. In order to extend the circle of participants as far as possible we advocate the use of natural language transmitted by email. We describe Cosma, a fully implemented German language server for existing appointment scheduling agent systems. Cosma can cope with multiple dialogues in parallel, and accounts for differences in dialogue behaviour between human and machine agents. NL coverage of the sublanguage is achieved through both corpusbased grammar development and the use of message extraction techniques.

Using the formalism of generalized phrase structure grammar (GF~SG) in an NL system (e.g. for machine translation (MT)) is promising since the modular structure of the formalism is very well suited to meet some particular needs of MT. However, it seems impossible to implement GPSG in its 1985 version straightforwardly. This would involve a vast overgeneration of structures as well as processes to filter out everything but the admissible tree(s). We therefore argue for a constructive version of GPSG where information is gathered in subsequent steps to produce syntactic structures. As a result, we consider it necessary to incorporate procedural aspects into the formalism in order to use it as a linguistic basis for NL parsing and generation. The paper discusses the major implications of such a modified view of GPSG. 1

DIRECT-INFO is a system for media monitoring currently applied to the field of sponsorship tracking. Significant parts of TV streams and electronic press feeds are automatically selected and subsequently monitored to find appearances of the name or logo of a sponsoring company in connection with the sponsored party. Basic features are automatically extracted from TV and press and thereafter fused to semantically meaningful results to support executive decision makers. Extracted features include detected logos, positive & negative mentions of a brand or product, multimodal video segmentation, speech-to-text transcripts and teletext subtitles, detected topics and genre classification. We first describe the technical workflow and architecture of the DIRECT-INFO system and then present its main innovations in the areas of logo detection, text analysis and fusion of results.

No abstract