Συμμετοχή της Comitech σε Ευρωπαϊκό Συνέδριο για e-government
H Comitech A.E. συμμετείχε σε εργασία που παρουσιάστηκε στο 12ο Ευρωπαϊκό Συνέδριο για e-government (ECEG 2012) που διεξήχθη στη Βαρκελώνη στις 14-15 Ιουνίου 2012. Ο τίτλος της εργασίας ήταν "MiCoGo, an Integrated System That Automatically Detects the Presence of Opinion in web Texts Regarding eCommerce and eGovernment" με συγγραφείς τους Γ. Στύλιο, Χ. Κατσή, Κ. Γλάρο, Β. Σιμάκη και Δ. Χριστοδουλάκη.
Abstract: The growth of Web 2.0 has facilitated interactive information sharing and interoperability on the World Wide Web, allowing users not only to retrieve information, but also the opportunity to interact or articulate their opinions on different topics. Platforms such as blogs, discussion forums, and various other types of social media provide citizens unprecedented power to share their experiences and opinions positive or negative, regarding any topic, including products or services. Therefore the Web can be considered as a big data repository and source consisting of a variety of data types as well as a large amount of unseen informative knowledge, which can be discovered via a wide range of data mining or machine learning paradigms. In this work, we present the concept and the initial results of a proposed system named MiCoGo. MiCoGo aims at providing an integrated set of tools that will automatically detect the presence of opinion in web texts regarding eCommerce and eGovernment. Research employs Semantic Web technologies, as well as linguistic knowledge and advanced machine learning and statistical techniques in order to: i) Separate accurately and promptly opinions about a product or service, or governmental decisions, into positive or negative, ii) weight the user’s comments according to the presence or not of arguments justifying their opinion about a certain service or product; iii) create a tool able to provide valuable, compact, but not deficient feedback about the users’ opinions regarding a company’s products or services; and v) make available to the participating enterprises a robust and objective tool for measuring customer satisfaction regarding a product or service. The proposed project will advance research in the specific domain and will provide valuable prototypes/linguistic tools. Moreover, MiCoGo will be applied to fora or blogs using Modern Greek, a language that imposes many specificities and difficulties in linguistic and resource level. The creation of such prototypes is innovative especially for the Greek language where the availability of linguistic tools is very limited.
Keywords: data mining; decision trees; eGovernment; support vector machines; web opinion mining