Design And Implementation Of An Intelligent Analysis System For Asset Evaluation Report Documents

With the advent of the Internet era,the number of electronic documents in enterprises has exploded,and the demand for their management and analysis has also increased.How to effectively analyze and manage massive unstructured documents,and discover valuable structured data from them,is an important and difficult problem in enterprise informatization construction.Asset evaluation report is a kind of financial document with high professionalism and important information content.It follows a specific chapter structure,but its content is complex,lengthy,and redundant.This paper aims at the tediousness and inefficiency of the information entry task of enterprise asset evaluation report,and designs intelligent methods and tools to achieve document parsing and element extraction of asset evaluation report.Firstly,this paper analyzes the structure and content characteristics of asset valuation reports and designs an intelligent document analysis scheme accordingly:1)Word document parsing,which parses the document content and chapter structure by traversing XML nodes to obtain chapter text blocks.2)Building a keyword dictionary,which trains a word vector model using the asset valuation report texts and generates synonyms through the model based on the manually collected keywords to expand the dictionary.3)extracting elements,segmenting the chapter text block according to the keyword dictionary to obtain the text to which the keywords belong;then extracting key-value elements and entity elements in the target text based on the defined element list using rule-based and named entity recognition respectively,and finally storing the extracted results as structured data.Secondly,based on the above scheme,a fully functional and wellinteractive intelligent analysis system for asset valuation report documents is designed and implemented using software engineering theories and methods,with the main functions of document management,document analysis,data management and user management.The system passed a thorough online test and ran well,achieving the designed target functions and meeting the actual business needs of the enterprise.The main innovations and contributions of this paper are in two aspects:First,it constructs an intelligent analysis scheme and a report element list for asset evaluation report documents,based on their features and requirements,providing an effective solution for the automatic processing of such documents.Second,it proposes a serial process scheme that first performs document parsing,then splits text according to keywords,and finally extracts elements using different methods,to tackle the problem of complex document content and improve the accuracy of information identification and extraction.