Language Technologies

back to overview

LiMoSINe - Linguistically Motivated Semantic aggregation engines

288024 - STREP

At a glance

FP7-ICT-2011-7 - Language technologies Duration: 36 months Start date: 1 November 2011 End date: 31 October 2014 Project officer: Pierre-Paul Sondag website Factsheet Workshop at LREC 2012

Challenge

This project aims at developing a multi-lingual web opinion mining system that enable the transition from a document-centric search paradigm to a semantic aggregation paradigm, hence getting directly the answer to a query instead of having to dig in the retrieved documents in order to find the answer.

Nowadays, Internet is used for everything from social interaction, playing and shopping to enjoying culture and sharing experiences, opinions or knowledge. The nature of online content has changed accordingly, from mostly edited material usually dealing with descriptions of facts to also include large volumes of user generated reports of personal or shared experiences, in many languages and often in multiple media. The projects partners aim the users to be offered search results and recommendations that are organized in semantically meaningful ways, based on entities, relations and key facts and sentiments associated with them. The languages covered within the project are Dutch, English, Italian and Spanish.

Objectives and Innovation

The vision of the LiMoSINe project is to develop semantic aggregation engines that provide information seekers with aggregates of information in response to a request, assembled from a variety of sources, in multiple languages, and organized and presented in meaningful patterns.
The key innovations of LiMoSINe are:
- the project focuses on online content with an emphasis on social media. This means that the tools need to be robust as the data can contain noise.
- the project will use accurate language processing tools plus semantic resources and design deep and precise semantic methods focused on identifying relations between entities in text.
- the project will advance the understanding in machine learning, in particular online learning - the project will perform large-scale semantic mining, inference and prediction using linguistic and extra-linguistic cues, including social network information and log files, thereby allowing us to link the web of documents and the web of data.
- the project will advance semantic personalization methods over streams of social media.

Target Group of the project

The technology developed can be used both by the general public using Yahoo Answer and Flickr (used cases), or by marketing professionals working on reputation management.

The result

The system will be integrated through web services. The project will exploit the research in three multilingual use cases: community-question answering Yahoo, online reputation management LLC and content annotation Flikr. The system will work in 4 languages Dutch, English, Italian and Spanish

Impact

The objective of LiMoSINe is to enhance the competitiveness of European industry in a multilingual digital market by providing language processing improved semantic mining techniques. The project will also enable collaboration and synergies between the natural language processing, the information extraction and the semantic web researcher communities.

Co-ordinator

Contact Person: Name: Prof. de Rijke Maarten Tel: +31 205255358 E-mail: m.derijke@uva.nl Organisation: UNIVERSITEIT VAN AMSTERDAM More»

 

 

 

 

back to overview