****************************************************** *********** EVALITA 2023: Call for tasks *********** ******************************************************
*EVALITA 2023 *is an initiative of AILC (Associazione Italiana di Linguistica Computazionale, *AILC* https://www.ai-lc.it/).
As in the previous editions (https://www.evalita.it/), EVALITA 2023 will be organized along a few selected tasks, which provide participants with opportunities to discuss and explore both emerging and traditional areas of *Natural Language Processing and Speech*. The participation is encouraged for teams working both in academic institutions and industrial organizations.
*TASK PROPOSAL SUBMISSION* Tasks proposals should be no longer than 4 pages and should include:
- task title and acronym; - names and affiliation of the organizers (minimum 2 organizers); - brief task description, including motivations and state of the art; - explanation of the international relevance of the task; - description and examples of the data, including information about their availability, development stage, and issues concerning privacy and data sensitivity. The examples are mandatory because they are intended to give potential participants an idea of what the task data will look like, how it’ll be formatted, etc. - expected number of participants and attendees; - names and contact information of the organizers.
/In submitting your proposal, please bear in mind that we encourage:/
- c*hallenging tasks* involving linguistic analysis, e.g., beyond “simple” classification problems; - tasks focused on multimodality, e.g., considering both textual and visual information; - tasks characterized by *different levels of complexity*, e.g., with a straightforward main subtask and one or more sophisticated additional subtasks; - the re-annotation/expansion of datasets from previous years with new annotation levels, and texts from publicly available corpora; - both new tasks and re-runs: for new tasks, organizers will have to specify in the proposal why it would attract a reasonable number of participants, and why it is needed; - application-oriented tasks, that is tasks that have a clearly defined end-user application showcasing; - *multilingual tasks*, i.e. with data both in Italian and in other languages; - *industrial tasks*, i.e. task with real data provided by companies.
The organizers of the accepted tasks should take care of planning, according to the scheduled deadlines (see below):
- the development and distribution of datasets needed for the contest, i.e. data for training and development, and data for testing; the scorer to be used to evaluate the submitted systems should be included in the release of development data; - the development of task guidelines, where all the instructions for the participation are made clear together with a detailed description of data and evaluation metrics applied for the evaluation of the participant results; - the collection of participants results; - the evaluation of participants results according to standard metrics and baseline(s); - the solicitation of participation and of submissions; - the reviewing process of the papers describing the participants approach and results (according to the template to be made available by the EVALITA 2023 chairs); - the production of a paper describing the task (according to the template to be made available by the EVALITA 2023 chairs).
**** Email your proposal in PDF format to evalita2023@gmail.com with "EVALITA 2023 TASK Proposal" as the subject line by the submission deadline: October 4th 2022. **** Please feel free to contact the EVALITA 2023 chairs at evalita2023@gmail.com in case of any questions or suggestions. * **Deadlines of the task proposal:* - October 4th 2022: submission of task proposals - October 18th 2022: notification of task proposal acceptance
*Timelines of EVALITA 2023: To be Announced Shortly*
*EVALITA 2023 CHAIRS* Mirko Lai (Università di Torino) Stefano Menini (Fondazione Bruno Kessler) Marco Polignano (Università di Bari Aldo Moro) Valentina Russo (Logogramma SRL) Rachele Sprugnoli (Università degli Studi di Parma) Giulia Venturi (Istituto di Linguistica Computazionale “A. Zampolli” - CNR)