TOV is the Duel’s platform that uses a technological architecture and data management process to offer a web service for the production of personalized traffic news based on the user’s position and profile. The bulletins are delivered in texts form and formatted to be read with Text-To-Speech (TTS) systems in a mobile device with GPS and Internet connection.
An intelligent selection of information
Traffic Over Voice platform captures and processes average speed data on the road network arcs (all motorways, extra-urban roads and most city streets) for which the Floating Car Data are available. TomTom traffic data is used via HD Flow service which is updated every two minutes.
TOV includes, integrates and standardizes (in terms of content consistency, geo-referencing and synchronization of event data) information – coming from traditional sources – which concern punctual events affecting traffic, such as roadworks, closures and accidents.
In order to generate a comprehensive network status information, TOV uses machine learning and artificial intelligence techniques to select and sort the most relevant news for each user, based on three factors: the statistics on recurring destinations on the road graphs, their profile, and the recent behavior. The system develops a specific format for voice messages publishing in natural language for safety and enjoyable access while driving. The platform uses a scalable, redundant infrastructure that works on virtual servers and cloud computing services provided by leading international companies (Google Cloud and Amazon Cloud Services).
The solution for complex technical processes
TOV’s architecture and process system solves complex problems such as:
- the definition of abstraction criteria and hierarchical road graphs and subgraphs;
- the definition of abstraction and extraction criteria for P.O.I. (points of interest) for the automatic creation of voice news with natural language;
- the application of machine learning techniques and simulation models for the identification of user’s destinations and traffic forecasts;
- the configuration and implementation of a complex information processing and distribution technology infrastructure for data volume and number of transactions to be processed in seconds.
A new system for the generation of customized bulletins
Traffic Over Voice offers a new bulletins generation system designed to use an external description of the selection criteria, grouping and ranking information written in a specifically designed language (Domain-Specific Language). Therefore, the system is able to manage both criteria used in mobile applications and those for web radio with the same software module. Through this approach, it is possible to customize the selection criteria, grouping and ranking for each service. The same type of approach has been followed for the natural language generation system: the software module accepts a description of the criteria for generating text from written data in a specific language, making it possible to add different languages and different bulletin styles without changing the bulletin generation system.
An open platform thanks to SDK for iOS and Android
Traffic Over Voice offers SDKs to facilitate the realization of third-party applications that use services on iOS and Android platforms. The SDKs allow client applications to use the platform services through high-level API interfaces that hide details of network protocols and data structures used for the session reducing drastically the implementation time and the possibility of errors.