A Two-Phase Bootstrap Approach to Facilitate Conversion from Text to Semantic Graph

Abstract

This paper presents a two-phase un-learnable approach to convert a text to its semantic graph. In the first phase: divide and conquer, the input text is divided into small pieces manageable by the available text to graph conversion tool (e.g.: Senna). This phase as such yields a collection of small subgraphs that don’t represent the entire text as a whole, however each individual small subgraph represents a corresponding piece of input text. In the second phase: focus attention, the underlying subgraphs are appended together by making use of a bootstrap algorithm to provide a strongly connected single graph that represents the entire input text. In the two phases, both SRL and RDF are considered and thoroughly explained. Accordingly, the corresponding two algorithms on divide and conquer and focus attention are bootstrapped (for both SRL and RDF), are evaluated and compared. The implementation of the such algorithms has indicated that this approach can be used with advantages of being simple, fast, straightforward and practical, which makes it attractive for those NLP researchers who are interested in converting texts to semantic graphs.