An Enhanced SVM Model for Implicit Aspect Identification in Sentiment Analysis

Opinion Mining or Sentiment Analysis (SA) is a key component of E-commerce applications where vast number reviews are generated by customers. SA operates on aspect level the views expressed specific product and have big influence customers’ choices businesses’ reputation. Aspect Based (ABSA) task categorizing text identifying sentiment attributed to it. Implicit Identification (IAI) subtask ABSA. This paper aims empirically investigate how external knowledge (e.g. WordNet) integrated into SVM model address some its intrinsic issues when dealing with classification. To achieve this research goal, we propose an approach improve Support Vector Machines (SVM) deal IAI. Using WordNet (WN) semantic relations, suggest enhancement for kernel computation. Experiments conducted three benchmark datasets products, laptops, restaurant reviews. The effects our examined analyzed according criteria: (i) function used, (ii) different experimental settings, (iii) behavior towards Overfitting Underfitting. finding work that integration experimentally proved be significantly helpful classification IAI especially addressing Underfitting considered as two main structural issues. empirical results demonstrate helps performance kernels under better