Projects and Research Activities

PITCH

PITCH (Prosodic Interaction, Talking, Care, Harmony) is an integrated system for evaluation of language and speech skills in subjects affected by psychic and neurological disorders. PITCHh uses audio data recorder during an interactive lesson between student and teacher.

PITCH aims at integrating different module typologies, for evaluating several independent aspects of the verbal behavior. PITCH concerns the following areas:

  • Basic linguistic skills (for spoken and written language)
  • Discourse skills
  • Dialogue skills
  • Crisis and discontinuity
  • Profiles, co-occurrences and correlations

DIKE

DIKE (Description of Interrogations by Knowledge Extraction), in the context of the DIKE-lab, deals with the analysis of pragmatic and linguistic aspects of dialogues and conversations, for forensic applications. DIKE will provide an architecture for the analysis of pragmatic and linguistic aspects of dialogues and conversations, allowing a complete analysis of speech. The approach considers textual and audio information; in particular, we will study the suprasegmental aspects of the vocal channel, and the prosodic / structural information of the text. The main applicative field will be in the forensic area.

In collaboration with Italian and Switzerland judicial authorities.

ATIS4all (finished)

Assistive Technologies and Inclusive Solutions for All (ATIS4all) is a European Thematic Network which main objective seeks to facilitate everyone’s access to the most suitable AT or accessibility device and service according to their needs, preferences and contextual characteristics (e.g. ICT solution, environment constraints, user device, language, etc.). For this purpose, ATIS4all will start and maintain an open, collaborative portal offering reliable information on AT and inclusive products and services, and cutting-edge technological trends. ATIS4all will contain Web 2.0 participation tools in order to encourage online discussion exchange of knowledge and expertise, and sharing of information among key actors and end users.

ArtDeco (finished)

ArtDeco is a nationally funded research project (MiUR FIRB 2005, project code RBNE05C3AH, 2006-2009). The project aims at developing techniques and methods for the diffusion of “networked enterprises” among the Italian Small and Medium Enterprises, testing the proposed approach in association with companies working in two test fields (food and textile/fashion) during all the phases of the project, from the initial definition of requirements to the development of prototypes/demonstrators, through a flexible and iterative process, where the technological development advances together with knowledge acquisition and with the refinement of network mechanisms.
To pursuit such results, the research consists of three, tightly coupled macro-goals:

  1. Organization and business models for networked enterprises;
  2. Acquisition, modeling and automatic management of business knowledge derived from heterogeneous sources, including natural language;
  3. Technological autonomic pervasive infrastructure, based on smart peripheral devices (Wireless Sensor Networks and RFID).

In particular, ARCSLab is involved in module 2. This module is complex since knowledge is often implicit and distributed in several different, not necessarily structured forms (from documents, to procedures, to working practices, to databases, …) and needs to be extracted and explicitly represented in order to be re-used and be considered the starting point for autonomous information management.
The research will define a methodology to extract and represent significant information for the networked enterprises, driven by an analysis of the domain ontologies. The task will focus on the extraction of information from documents, process actions and from other different types of repositories (structured and not- structured), and on its formalization to make it available in the most appropriate way.

UbiCampus (finished)

The Ubiquitous Campus UbiCampus project aims at:

  • Permitting indoor-outdoor localization and navigation by means of an advanced stochastic positioning model, which measures the power of WiFi signals and exploits the accelerometer/gyroscope/GPS sensors of modern smartphones.
  • Raching a complete accessibility of the campus, by means of a detailed description of the geometric and accessibility-related attributes of the campus.
  • Obtaining personalized services based on a specific profile. in particular, we developed ICF*, a subset of the original WHO ICF specification, extended with new, technology-oriented attributes.
  • Permitting automatic, personalized path calculations, exploiting the localization system, the description of the campus, and the user profile.

Campus Tools (CATS) (finished)

The Campus Tools (CATS) project aims at facilitating multimodal access to lectures. The project involves three Universities: Politecnico di Milano, Università  di Modena e Reggio Emilia, Università  di Milano Bicocca. In particular the ARCSLab is developing:

  • A hardware/software solution to facilitate lip-reading for deaf or hard of hearing students. The system uses a microcamera and a microphone, weared by the teacher, to send an audio/video stream of lip movements and speech; the stream also includes automatic captioning by means of an ASR application. Students attending the class receive the audio/video/textual stream on their personal device (laptop, smartphone, etc.) via WiFi.
  • A software solution to facilitate notetaking; the objective of our work is the design and development of a collection of software applications that allow convenient note-taking during slide-based classes. Such applications permit students to receive on their Tablet-PCs the slides presented by teachers, in real time. Students can edit objects contained in the slides, as well as add their own pen-based notes. The resulting notes, thus, contain text, images, video, sounds, and hand-written parts (sketches or phrases).
  • A tool able to recognize relevant concepts of texts, and relationships among them. Such information is used to generate summaries and so-called mental maps, used by persons with dyslexia. The system uses a-priori knowledge about the domain, by meas of an ontology about actors and their actions; verb categorization and frames permit to disambiguate the meaning of words and map them onto the related ontology concepts; finally, inference rules permit to extend the original domain ontology.
  • A specialized spell checker and predictor for people with dyslexia, based on an advanced stochastic model and a well-designed user interface.