A new version of the Java Machine Learning platform KeLP developed within the SAG group has been released.
Many fixes and new functionalities are included in this version. Among them, an efficient and full version of the Smoothed Partial Tree Kernel is for the first time available to the public.
Check out this new version from our repositories. Soon we will upload new versions of the documentation pages, while API Javadoc is already available.
Your suggestions will be very precious for us, so download and try KeLP 1.1.0!
The paper Acquiring a Large Scale Polarity Lexicon through Unsupervised Distributional Methods by Castellucci, Croce and Basili has been accepted at the 20th NLDB Conference.
Reveal Srl, a SAG Spin-Off, is one of the finalists at UNIRSI PER L’IMPRESA startup prize. The challenge is organized by UNINDUSTRIA – Unione degli Industriali e delle Imprese in Lazio.
SAG team (Castellucci, Croce, De Cao and Basili) participated to the SENTIment POLarity Classification (SENTIPOLC) task within the context of EVALITA 2014.
SAG team obtained very good results, it ranked between 1st and 4th position in the different sub-tasks, and won 3 of the 6 evaluation carried out by the task organizers. Moreover, SAG team won the pilot task of irony detection in Italian.
See you at EVALITA in Pisa!
Two papers have been accepted at the The first Italian Computational Linguistics Conference. The work Towards Compositional Tree Kernels by Annesi, Croce and Basili and A context based model for Twitter Sentiment Analysis in Italian by Vanzo, Castellucci, Croce and Basili will be presented at the conference in Pisa in December.
The paper Semantic Compositionality in Tree Kernels by Basii, Annesi and Croce has been accepted for publication on the ACM International Conference on Information and Knowledge Management (CIKM 2014). See you at Shangai!
DiEM project will be at the SMAU 2014 in Milan (22-24 October 2014).
DiEM system has been designed to query heterogeneous databases using natural language. Both written and oral interactions between a user and the system are allowed. For more details, check the SMAU page.
The paper Effective and Robust Natural Language Understanding for Human Robot Interaction from Bastianelli, Castellucci, Croce, Basili and Nardi has been accepted for publication at the European Conference on Artificial Intelligence (ECAI 2014).
We are proud to announce that the paper A context-based model for Sentiment Analysis in Twitter by Vanzo, Croce and Basili has been awarded with the IBM Best Paper Award at the Coling 2014 conference in Dublin.
SAG team (Castellucci, Filice, Croce, Basili) participated to the Aspect Based Sentiment Analysis task!
We used Structured Learning paradigms to tackle the task, check it in the paper UNITOR: Aspect Based Sentiment Analysis with Structured Learning. We obtained promising results, we ranked between 2nd and 8th position in the different tasks!