Profile cover photo
Profile photo
GLiCom
5 followers -
GLiCom -- Grup de Recerca en Lingüística Computacional
GLiCom -- Grup de Recerca en Lingüística Computacional

5 followers
About
GLiCom's posts

Post has attachment
Event Understanding through Multimodal Social Stream Interpretation (EUMSSI - FP7-ICT-2013-10-611057) 

EUMSSI (http://www.eumssi.eu/) aims at developing technologies for identifying and aggregating data presented as unstructured information in sources of very different nature (text, video, audio and social media), including media from online news providers and broadcasters as well as from social sites (e.g. Twitter, YouTube). This is accomplished thanks to the integration in a cross-modal semantic representation framework of state-of-the-art technologies from text and audio-visul analysis, audio-visual analysis. The multimodal interpretation EUMSSI platform, in an optimized process chain, analyses a vast amount of multimedia content and enriches it with additional metadata layers. These metadata layers provide a multi-facetted access of comprehensive information and can be used for further analyses or enrichment. A core idea is that the aggregation of information is carried out in an interactive manner, so that the metadata resulting from one media helps reinforce the analysis results from other media. The resulting platform, through the use of innovative visualization techniques will allow an interactive exploration of the enriched data helping discover new, implicit information.

Post has attachment
GLiCom Research Seminar

Speakers: Maite Melero and Jens Grivolla (Universitat Pompeu Fabra)

Event Understanding through Multimodal Social Stream Interpretation (EUMSSI - FP7-ICT-2013-10-611057) 

EUMSSI (http://www.eumssi.eu/) aims at developing technologies for identifying and aggregating data presented as unstructured information in sources of very different nature (text, video, audio and social media), including media from online news providers and broadcasters as well as from social sites (e.g. Twitter, YouTube). This is accomplished thanks to the integration in a cross-modal semantic representation framework of state-of-the-art technologies from text and audio-visul analysis, audio-visual analysis. The multimodal interpretation EUMSSI platform, in an optimized process chain, analyses a vast amount of multimedia content and enriches it with additional metadata layers. These metadata layers provide a multi-facetted access of comprehensive information and can be used for further analyses or enrichment. A core idea is that the aggregation of information is carried out in an interactive manner, so that the metadata resulting from one media helps reinforce the analysis results from other media. The resulting platform, through the use of innovative visualization techniques will allow an interactive exploration of the enriched data helping discover new, implicit information.

When: 6 may 2015, 12.00-13.00 

Where: room 52.415, Edifici Roc Boronat (Campus de la Comunicació) 

Post has attachment
First public presentation of the EUMSSI platform demonstrator at the "International Workshop on Knowledge Extraction and Semantic Annotation" (http://www.iaria.org/conferences2015/KESA.html) in Barcelona.

This demonstrator showcases the ongoing work in the EUMSSI project and will lead to two prototype systems: an assistant for journalists to help discover content and trends, and a second screen application to provide home users with related content when watching broadcast news.

Post has attachment
Speaker: Horacio Saggion (Universitat Pompeu Fabra)
"Practical Multilingual Text Simplification"

Post has attachment
GLiCom Research Seminar

Speaker: Horacio Saggion (Universitat Pompeu Fabra)

Practical Multilingual Text Simplification  

Text simplification aims at transforming a complex text into a more readable and understandable version which conveys approximately the same content. The transformation usually implies the replacement of words which are difficult to understand by easy synonyms and the re-writing of long and complicated sentences into shorter ones. Automatic text simplification develops algorithms for the automation of this process and although the first works in this area appeared in the nineties, the research in this field has nowadays intensified. We will present an overview of the natural language processing techniques we have used for the development of simplification technology for Spanish and English in the Simplext and Able-to-Include projects.

When: 4 march 2015, 12.00-13.00 

Where: room 52.415, Roc Boronat Building (Campus de la Comunicació) 

Post has attachment
Speaker: Carlos Hernandez Mena (Universidad Nacional Autónoma de México)
"A novel linguistic software infrastructure for speech recognition systems creation, in Mexican Spanish"

Post has attachment
GLiCom Research Seminar

Speaker: Carlos Hernandez Mena (Universidad Nacional Autónoma de México)

A novel linguistic software infrastructure for speech recognition systems creation, in Mexican Spanish  

Nowadays, the field of automatic speech recognition (ASR) is a mix of several other fields like the computer sciences, acoustics, pattern recognition, signal processing and so many others. One of these disciplines which is especially important to the ASR systems is linguistics. Linguistics is always present in the training corpus that helps the system to "learn" the speech sounds and also in the pronouncing dictionary that helps the system to determine how a certain word is pronounced. 

In the following presentation, a novel Mexican Spanish corpus designed to train ASR systems will be presented, an old computational phonetic alphabet called "Mexbet", created for the Spanish of Central Mexico will be explored, a phonetizer tool based on Mexbet will be shown and a set of tools designed to create  pronouncing dictionaries will be reviewed. It is important to mention that every software tool and language resource presented here is open-source and available on the internet at absolutely no cost.

When: 18 February 2015, 12.00-13.00 

Where: room 52.735, Roc Boronat Building(Campus de la Comunicació) 

Post has attachment

Post has attachment
Joint workshop of the Multisensor and +eumssi projects @ UPF

February 11 in room 55.309 (Tanger building - Poblenou campus).

Post has shared content
Press release by one of our partners in the +eumssi project.
Along with key European institutions in the field of research and technological development, such as Pompeu Fabra University, German public broadcaster Deutsche Welle, IDIAP Research Institute from Switzerland or the Leibniz University of Hannover, VSN is working on the development of a new technology to help journalists in their daily work.
Wait while more posts are being loaded