Digital document analysis is one where software analysts review documents for assessing an appraisal theme. Digital document analysis can be utilized for obtaining available documents in order to extract relevant data. Most of the research work focuses on a semi-supervised based framework for better parsing performance and traditional statistical setting. However, an inappropriate selection during digital documents analysis may lead to entire process being falsified there by reducing the overall accuracy. To address this issue, in our work, a novel method called, Weighted Score Convolutional Network and Arc-factored Graph-based Dependency Parsing (WSCN-AGDP) is proposed. WSCN-AGDP is split into two sections. First section is concerned with the extraction of relevant features (i.e., words from sentences) by employing Stouffer’s Weighted Score-based Convolutional Neural Network model. In the second section, using the extracted features, Graph-based Dependency Parsing is performed by utilizing Spearman Correlated Arc-Factored model. Four indices were calculated namely, digital document parsing time, parsing overhead, false positive rate and precision are being used to quantitatively assess and rate the algorithms. Different document sizes acquired from Reuters-21578 dataset are considered. Experiments have been conducted to analyze the methods.

Digital document analysis is one where software analysts review documents for assessing an appraisal theme. Digital document analysis can be utilized for obtaining available documents in order to extract relevant data. Most of the research work focuses on a semi-supervised based framework for better parsing performance and traditional statistical setting. However, an inappropriate selection during digital documents analysis may lead to entire process being falsified there by reducing the overall accuracy. To address this issue, in our work, a novel method called, Weighted Score Convolutional Network and Arc-factored Graph-based Dependency Parsing (WSCN-AGDP) is proposed. WSCN-AGDP is split into two sections. First section is concerned with the extraction of relevant features (i.e., words from sentences) by employing Stouffer’s Weighted Score-based Convolutional Neural Network model. In the second section, using the extracted features, Graph-based Dependency Parsing is performed by utilizing Spearman Correlated Arc-Factored model. Four indices were calculated namely, digital document parsing time, parsing overhead, false positive rate and precision are being used to quantitatively assess and rate the algorithms. Different document sizes acquired from Reuters-21578 dataset are considered. Experiments have been conducted to analyze the methods.