A Textual Entailment System using Anaphora Resolution

The note describes the Recognizing Textual Entailment (RTE) system developed at the Computer Science and Engineering Department, Jadavpur University, India. In this competition, we have participated and submitted the results in the RTE-7 Main Task (3 runs), Novelty Task (3 runs) and RTE-7 KBP Validation task (2 unique runs for generic task and 2 unique runs for tailored task). For the RTE7 Main and Novelty Tasks, the systems are based on pre-processing task which includes Anaphora Resolution using JavaRAP tool then the system is the composition of Lexical Entailment module, Syntactic Entailment module, Chunk module and Named Entity module. For the RTE-7 Main task test set, the following micro-average results were obtained for Run 1: F-Score 29.81, Run 2: F-Score 30.47 and Run 3: F-score 29.90. For the RTE-7 Novelty task test set, the following micro-average results were obtained for Run 1: Novelty Evaluation F-Score 86.26 and Justification Evaluation F-Score 20.02, Run 2: Novelty Evaluation F-Score 78.49 and Justification Evaluation F-Score 26.56 and Run 3: Novelty Evaluation F-score 73.94 and Justification Evaluation F-Score 25.55 were obtained. The RTE-7 KBP Validation Task is based on the assumption that extracted slot filler is correct if and only if the supporting document entails a hypothesis created on the basis of the slot filler. In RTE KBP, we participated for generic task and tailored task. For the RTE-7 KBP Validation task test set for Generic Task, micro-average results for Run 1: F-Score 0.148 and Run 2: F-Score 0.1902 were obtained. For RTE-7 KBP test set for Tailored Task, micro-average results for Run 1: F-Score 0.1813, Run 2: F-Score and 0.1834 were obtained.