Generating natural language traffic information from structured traffic data.

Traveler Information Services (TIS) play an increasing role in ITS and therefore the center backbone carrying the underlying real-time information grows steadily. New technologies and standards provide more flexibility and detailed models carry structured data at increasingly high complexity compared to simple ALERT-C coding. But these technical improvements make generating natural text more complicated. Textual representations are required for many 'old' (radio, WWW) but also 'new' (WAP, PDA) services. This paper describes a generic concept how this problem can be solved and gives experiences from first implementation in Traffic Information Centers in Germany.