A Position Aware Decay Weighted Network for Aspect based Sentiment Analysis

Aspect Based Sentiment Analysis (ABSA) is the task of identifying sentiment\npolarity of a text given another text segment or aspect. In ABSA, a text can\nhave multiple sentiments depending upon each aspect. Aspect Term Sentiment\nAnalysis (ATSA) is a subtask of ABSA, in which aspect terms are contained\nwithin the given sentence. Most of the existing approaches proposed for ATSA,\nincorporate aspect information through a different subnetwork thereby\noverlooking the advantage of aspect terms' presence within the sentence. In\nthis paper, we propose a model that leverages the positional information of the\naspect. The proposed model introduces a decay mechanism based on position. This\ndecay function mandates the contribution of input words for ABSA. The\ncontribution of a word declines as farther it is positioned from the aspect\nterms in the sentence. The performance is measured on two standard datasets\nfrom SemEval 2014 Task 4. In comparison with recent architectures, the\neffectiveness of the proposed model is demonstrated.\n