BTPK-based interpretable method for NER tasks based on Talmudic Public Announcement Logic

TL;DR

This paper introduces BTPK, an interpretable logic-based method for NER that reveals the internal decision process of neural models, enhancing transparency and understanding of how context influences recognition.

Contribution

The paper presents a novel BTPK model based on Talmudic Public Announcement Logic that explains neural NER models' inner workings and captures semantic context.

Findings

BTPK reveals the decision logic of BRNNs in NER tasks.

The explanations from BTPK show how context influences recognition.

BTPK effectively captures semantic information in input sentences.

Abstract

As one of the basic tasks in natural language processing (NLP), named entity recognition (NER) is an important basic tool for downstream tasks of NLP, such as information extraction, syntactic analysis, machine translation and so on. The internal operation logic of current name entity recognition model is black-box to the user, so the user has no basis to determine which name entity makes more sense. Therefore, a user-friendly explainable recognition process would be very useful for many people. In this paper, we propose a novel interpretable method, BTPK (Binary Talmudic Public Announcement Logic model), to help users understand the internal recognition logic of the name entity recognition tasks based on Talmudic Public Announcement Logic. BTPK model can also capture the semantic information in the input sentences, that is, the context dependency of the sentence. We observed the public announcement of BTPK presents the inner decision logic of BRNNs, and the explanations obtained from a BTPK model show us how BRNNs essentially handle NER tasks.