TELEPARES

An ever growing number of web users employ microtext-based social networks to share their opinions and experiences on different products, services or persons. This situation has sparked a surge of interest in developing platforms that can effectively exploit the untapped potential of this information, so as to learn the point of view of a multilingual user base on a large variety of subjects. Sentiment analysis, also know as opinion mining, is a recent research field which focuses on automatically finding out whether a text is opinionated or not, whether the polarity or sentiment expressed in it is positive, negative or mixed; and on automatically extracting the author's perception on particular aspects of a subject. However, existing solutions are limited by their scant use of language technologies, as they perform a shallow processing without taking into account syntactic relations between words and their semantic roles. This hampers their ability to understand texts that are already intrinsically difficult due to their brevity. Furthermore, the majority of these tools assume English as their base language, resulting on a comparative advantage for users, institutions and companies from English-speaking countries.
Although microtexts present some specific lexical and syntactic properties that differ from those of standard text, certain basic aspects of language must be respected so that they are intelligible. In this project, we propose to exploit this fact in order to improve the linguistic support for processing microtexts in our natural sphere of interest: the Spanish and Galician languages. This is especially relevant in the light of recent European reports which have identified a shortage in resources for language technology support on these languages, making special emphasis on the lack of syntactic resources. The achievement of this goal will require to combine the knowledge and experience of the participating teams in the fields of Computational Linguistics, Information Retrieval and Knowledge Acquisition.
The final goal is to develop and effective opinion analysis system working on Spanish and Galician for microtext-based social networks. To do so, it will be necessary to improve the performance of current parsing and analysis techniques on standard text, to design mechanisms so that models and methods effective for analyzing standard language can be adapted to microtexts, and to project effective models, methods and resources across languages.

<p>An ever growing number of web users employ microtext-based social networks to share their opinions and experiences on different products, services or persons. This situation has sparked a surge of interest in developing platforms that can effectively exploit the untapped potential of this information, so as to learn the point of view of a multilingual user base on a large variety of subjects. Sentiment analysis, also know as opinion mining, is a recent research field which focuses on automatically finding out whether a text is opinionated or not, whether the polarity or sentiment expressed in it is positive, negative or mixed; and on automatically extracting the author's perception on particular aspects of a subject. However, existing solutions are limited by their scant use of language technologies, as they perform a shallow processing without taking into account syntactic relations between words and their semantic roles. This hampers their ability to understand texts that are already intrinsically difficult due to their brevity. Furthermore, the majority of these tools assume English as their base language, resulting on a comparative advantage for users, institutions and companies from English-speaking countries.<br />Although microtexts present some specific lexical and syntactic properties that differ from those of standard text, certain basic aspects of language must be respected so that they are intelligible. In this project, we propose to exploit this fact in order to improve the linguistic support for processing microtexts in our natural sphere of interest: the Spanish and Galician languages. This is especially relevant in the light of recent European reports which have identified a shortage in resources for language technology support on these languages, making special emphasis on the lack of syntactic resources. The achievement of this goal will require to combine the knowledge and experience of the participating teams in the fields of Computational Linguistics, Information Retrieval and Knowledge Acquisition.<br />The final goal is to develop and effective opinion analysis system working on Spanish and Galician for microtext-based social networks. To do so, it will be necessary to improve the performance of current parsing and analysis techniques on standard text, to design mechanisms so that models and methods effective for analyzing standard language can be adapted to microtexts, and to project effective models, methods and resources across languages.</p> - Pablo Gamallo Otero