The QALL-ME Project (Question Answering Learning technologies in a multiLingual and Multimodal Environment) is an EU Commission funded initiative envolving seven institutions from four member countries (Italy, United Kingdom, Spain and Germany) to establish a shared infrastructure for multilingual and multimodal open domain Question Answering for mobile phones.QALL-ME will provide innovative solutions to automatically manage huge amounts of multilingual information coming from different source data types, allow users for an easy and natural means of interaction, based on natural language, through their mobile telephone or PDA, present users with rich and reliable information anywhere and at any time.

There is enormous market potential for achievements in the directions pursued in the QALL-ME project, as the exponential growth of requests to call centres implies. High precision QA services will dramatically reduce the time required of human personnel to provide answers. As a consequence, the ability to automatically address even a small proportion of such information traffic will offer new revenue opportunities for those companies working in telecommunications and in the web-based information services scenario. From a commercial point of view, high-performance technologies for the extraction, integration and fusion of information from semi-structured and structured heterogeneous data pools are essential factors for future intelligent business models in the area of e-commerce, but also for e-learning applications.

The general objective of QALL-ME is to establish a shared infrastructure for multilingual and multimodal open domain Question Answering for mobile phones. QA takes a question in natural language and returns an answer from a collection of information sources (e.g. documents, databases). In contrast to the technologies behind today's web search engines, the goal of QA is not to return a document which contains the answer, as in the case of information retrieval, but the actual sequence of words which constitutes the answer. The potential of open domain QA will be experimented with and evaluated in the context of mobile applications for information seeking, a multimodal scenario which includes spontaneous speech as input and the integration of textual answers with maps, images, and short videos as output. In this context QA fits well with the need for both short questions and short answers, due to the limited capacity of small screens in mobile devices (e.g. smart phones, palm tops).

The QALL-ME Consortium is composed of seven institutions from four member countries (Italy, United Kingdom, Spain and Germany). Four of the participants are academic institutions (ITC-irst, University of Wolverhampton, University of Alicante, and DFKI), while the others are industrial partners (Comdata, Ubiest, and Waycom).

The specific research objectives of the project include state-of-art advancements in the complexity of the questions handled by the system(e.g. how questions); the development of a web-based architecture for cross-language QA (i.e. question in one language, answer in a different language); the realization of real time QA systems for concrete applications; the integration of the temporal and spatial context both for question interpretation and for answer extraction; the development of a robust framework for applying minimally supervised machine learning algorithms to QA tasks; and the integration of mature technologies for automatic speech recognition within the open domain question answering framework.

UbiEst will be in charge of putting its mapping and location technologies experience into the realization of innovative services, in order to guarantee that the QALL-ME technology meets all the market requirements, both from a user-oriented perspective and from the business perspective.

In particular UbiEst will provide mapping and geocoding capabilities for innovative local searches: QALL-ME integrates the results of its component-level research activity with existing multimodal functionalities considering different source data types and different output formats (i.e. maps, images, short videos) as possible ways to present users with the sought information.

This will allow the system to enrich a textual answer with additional information (e.g. combining the address of a restaurant with an image of the building in which it is located), or to provide the user with different output types more suitable as answers to his question (e.g. a map of the city, a movie trailer).Input questions will be automatically classified according to the request they pose and appropriate answer modalities (e.g. text only, text + video, text + map + route, etc.) will be selected. The classification is based on machine learning algorithms, trained on the QALL-ME benchmark and integrated with the system's "episodic memory" (i.e. records of successful interactions).

See http://www.ubiest.com/