Sorry for cross-posting
Spanish version below
********************************************************************
TASS-2017: Workshop on Semantic Analysis at SEPLN
Sep 19th, 2017
Murcia, Spain
http://www.sepln.org/workshops/tass/2017
********************************************************************
Call for participation
The workshop and shared task Sentiment Analysis at SEPLN (TASS) has been celebrated at the International Congress of the Spanish Society of Natural Language Processing (SEPLN) since 2012. TASS was the first shared task focused on sentiment analysis in Twitter in Spanish. Spanish is the second language used in Facebook and Twitter, so it is very important the study of specific methods for the processing of Spanish and the creation of resources for sentiment analysis in Spanish. Therefore, the initial aim of TASS was the furtherance of the research on sentiment analysis in Spanish with a special interest on the language used in Twitter.
Although sentiment analysis is still an open problem, the Organization Committee wants to foster the research on other tasks related to the processing of the semantics of texts written in Spanish. Consequently, the name of the workshop/shared task has been changed to Workshop on Semantic Analysis at SEPLN (TASS).
TASS 2017 will be held the 19th of September and it is co-located with the International Conference of the Spanish Society for Natural Language Processing, which will take place at Murcia, Spain.
The Organization Committee of TASS 2017 proposes two evaluation tasks of polarity classification systems of opinions written in Spanish. Specifically, the tasks are the following:
*** Task 1: Sentiment Analysis at Document Level ***
The aim of the task is the evaluation of polarity classification systems at document level (tweet), which have to classify the intensity of the opinion expressed in intensity range of 4 steps (P, NEU, N, NONE). The novelty of TASS-2017 is the proposal of a new dataset for training and test, which is called InterTASS and its features are described in the web page of TASS.
The submitted systems must be trained on the InterTASS training set, and they must be evaluated on the test set of InterTASS and the General Corpus of TASS (see previous editions). Participants are expected to submit three experiments per each evaluation set, so each participant team can submit a maximum of 6 systems.
The evaluation measures will be the macro-averaged version of Precision, Recall and F1, and Accuracy. The systems will be ranked by the Macro-F1 and Accuracy measures.
*** Task 2: Sentiment Analysis at Aspect Level ***
The aim of the second task is the evaluation of polarity classification systems at aspect level. Those systems have to work with tweets from two different domains: politics (STOMPOL corpus) and media events (Social-TV corpus).
Participants are expected to submit three experiments per corpus, so 6 maximum number of experiments in total. The evaluation measures will be the macro-averaged version of Precision, Recall and F1, and Accuracy. The systems will be ranked by Macro-F1 and Accuracy measures.
Submission of papers
The participants of each of the two tasks are invited to submit a description paper of the systems submitted. All the papers will be reviewed by the program committee, and the accepted ones will be published in the proceedings of TASS 2017. The proceedings will be published in CEUR.org as the proceedings of the previous editions (2015 and 2016). The guidelines for the papers are published in the web page of TASS.
Important dates
Release of the training data May 1, 2017
Release of the test data June 20, 2017
Registration June 30, 2017
Submission of systems July 1, 2017
Submission of papers July 15, 2017
Decision notification July 31, 2017
Camera-ready submission August 31, 2017
Publication September 15, 2017
TASS September 19, 2017
Organization Commitee
Julio Villena-Román Meaning Cloud, Spain
Miguel Ángel García Cumbreras University of Jaén, Spain
Eugenio Martínez Cámara Technische Universität Darmstadt, Germany
Manuel Carlos Díaz Galiano University of Jaén, Spain
Manuel García Vega University of Jaén, Spain
Program Committee
Edgar Casasola Murillo Universidad de Costa Rica, Costa Rica
Fermín Cruz Mata Universidad de Sevilla, Spain
Yoan Gutiérrez Vázquez Universidad de Alicante, Spain
Lluís F. Hurtado Universidad Politécnica de Valencia, Spain
Salud María Jiménez Zafra University of Jaén, Spain
Mª. Teresa Martín Valdivia University of Jaén, Spain
Manuel Montes Gómez Instituto Nacional de Astrofísica, Óptica y Electrónica, Mexico
Antonio Moreno Ortíz University of Málaga, Spain
Preslav Nakov Qatar Computing Research Institute, Qatar
José Manuel Perea Ortega Universidad de Extremadura, Spain
Ferrán Pla Universidad Politécnica de Valencia, Spain
Sara Rosenthal IBM Research, U.S.A.
Maite Taboada Simon Fraser University, Canada
L. Alfonso Ureña López University of Jaén, Spain
********************************************************************
TASS-2017: Taller de Análisis Semántico en la SEPLN
19 de septiembre de 2017
Murcia, España
http://www.sepln.org/workshops/tass/2017
********************************************************************
Convocatoria de participación
El taller de evaluación sobre Análisis de Sentimientos en la SEPLN (TASS) se ha venido celebrando en el Congreso Internacional de la Sociedad Española de Procesamiento del Lenguaje Natural (SEPLN) desde el año 2012. TASS fue el primer taller de evaluación de sistemas de análisis de opiniones escritas en español. El español es la segunda lengua más usada en Facebook y en Twitter, por lo que es capital el estudio de técnicas específicas para el procesamiento del español, así como la generación de recursos lingüísticos propios para la tarea de análisis de opiniones en español. Por tanto, el objetivo original de TASS fue el fomento y promoción de la investigación en análisis de opiniones en español con especial atención al lenguaje que se usa en Twitter.
Aunque todavía queda mucho por hacer en la tarea de análisis de opiniones, el Comité Organizador pretende promover la investigación de tareas de análisis semántico sobre textos escritos en español. Por consiguiente, el nombre de TASS cambia este año a Taller de Análisis Semántico en la SEPLN (TASS).
La edición de 2017 se celebrará el 19 de septiembre conjuntamente con el Congreso Internacional de la Sociedad Española de Procesamiento del Lenguaje Natural, que en esta ocasión se organiza en Murcia, España.
El comité organizador de TASS 2017 propone dos tareas de evaluación de sistemas de clasificación de opiniones escritas en español. En concreto, las dos tareas son las siguientes:
*** Tarea 1: Análisis de Opiniones a nivel de documento ***
El objetivo es el desarrollo de un sistema de clasificación de opiniones a nivel de documento (tuit), el cual debe determinar la intensidad de la opinión expresada en una escala de 4 (P, NEU, N, NONE) niveles. La novedad de la edición de 2017 radica en el uso de una nueva colección de entrenamiento y evaluación, que se llama InterTASS y cuyas características se encuentran descritas en la web de TASS.
Los sistemas que se propongan tienen que haber sido entrenados única y exclusivamente con el conjunto de entrenamiento del corpus InterTASS, y evaluados con el conjunto de evaluación (test) de InterTASS y del Corpus General de TASS (ver ediciones anteriores). Los participantes podrán enviar un máximo de tres sistemas por cada corpus de evaluación (InterTASS y Corpus General), de manera que como máximo un mismo equipo puede proponer 6 sistemas.
Las medidas de evaluación que se utilizarán son la media aritmética (macro-average) de la Precision, Recall y F1 por clase, y el Accuracy. Los sistemas se ordenarán atendiendo al valor de macro-F1 y de Accuracy.
*** Tarea 2: Análisis de Opiniones a nivel de aspecto ***
El objetivo de la segunda tarea es la evaluación de sistemas de clasificación de opiniones a nivel de aspecto. Para ello, los participantes tendrán que desarrollar sistemas de clasificación de opiniones a nivel de aspecto adaptados a dos dominios diferentes: política (corpus STOMPOL) y evento en televisión (corpus Social-TV).
Los participantes podrán enviar tres sistemas diferentes por corpus. Las medidas de evaluación que se utilizarán son la media aritmética (macro-average) de la Precision, Recall y F1 por clase, y el Accuracy. Los sistemas se ordenarán atendiendo al valor de Macro-F1 y de Accuracy.
Envío de artículos
Los participantes en cualquiera de las dos tareas podrán enviar un artículo de descripción de su sistema. Si los artículos superan la revisión por parte de los miembros del Comité de Programa, serán publicados en las actas de TASS 2017. Las actas se publicarán en la plataforma CEUR.org, al igual que se hizo en las ediciones de 2015 y 2016. La guía de estilo de los artículos está publicada en la web de TASS.
Fechas relevantes
Publicación de conjunto de datos de entrenamiento 1 de mayo de 2017
Publicación de conjunto de datos de evaluación 20 de junio de 2017
Registro 30 de junio de 2017
Envío de sistemas y evaluación 1 de julio de 2017
Envío de artículos 15 de julio de 2017
Notificación de revisión 31 de julio de 2017
Envío versión final del artículo 31 de agosto de 2017
Publicación 15 de septiembre de 2017
TASS 19 de septiembre de 2017
Comité Organizador
Julio Villena-Román Meaning Cloud, España
Miguel Ángel García Cumbreras Universidad de Jaén, España
Eugenio Martínez Cámara Technische Universität Darmstadt, Alemania
Manuel Carlos Díaz Galiano Universidad de Jaén, España
Manuel García Vega Universidad de Jaén, España
Program Committee
Edgar Casasola Murillo Universidad de Costa Rica, Costa Rica
Fermín Cruz Mata Universidad de Sevilla, España
Yoan Gutiérrez Vázquez Universidad de Alicante, España
Lluís F. Hurtado Universidad Politécnica de Valencia, España
Salud María Jiménez Zafra Universidad de Jaén, España
Mª. Teresa Martín Valdivia Universidad de Jaén, España
Manuel Montes Gómez Instituto Nacional de Astrofísica, Óptica y Electrónica, México
Antonio Moreno Ortíz Universidad de Málaga, España
Preslav Nakov Qatar Computing Research Institute, Catar
José Manuel Perea Ortega Universidad de Extremadura, España
Ferrán Pla Universidad Politécnica de Valencia, España
Sara Rosenthal IBM Research, U.S.A.
Maite Taboada Simon Fraser University, Canadá
L. Alfonso Ureña López Universidad de Jaén, España