Semantic based classification of trademark cases using Naïve Bayes & SVM

Abstract

Trademarks are signs that are utilized in exchange to recognize items or administrations. They have turned out to be intangible intellectual property (IP) resources that enable merchandise or administrations to be effectively perceived by customers. The quantity of trademarks enrolled and utilized every year in the commercial center demonstrates an upward pattern with no huge indication of declining. Many trademark cases have been documented till to date. Consequently, it has been turning into a challenge to classifications the trademark case in view of context orientation. Old classifiers don't include semantics, implies they don't think about semantic likeness between sentences. Conventional classification scheme is restricted to detecting patterns within the used terminology only, which excludes conceptual patterns as well as any semantically related words. Along these lines, semantic based classification is proposed. It is in this manner conceivable to increase better outcomes over different spaces by using an outside semantic thesaurus like WordNet that characterizes an upper-level of connections among the vast majority of the terms in the testing information. We have performed experiments on the data set of "Pakistan legal Trademark case studies" by using machine learning algorithm Such as Naïve Bayes and SVM with the combination of domain Ontology such as WordNet. We have performed comparison with old classifiers (NB, SVM) but our approach achieves better accuracy result of 77% by using NB and 81% by using SVM.