A knowledge management method for non-performing assets management based on natural language processing technology

A technology of natural language processing and business management, applied in the financial field, can solve problems such as difficult access, low degree of market participation, and high entry threshold for non-performing asset management business, and achieve the effect of improving accuracy and avoiding the disaster of dimensionality

Active Publication Date: 2021-08-13
华融融通(北京)科技有限公司
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] Due to the particularity of the industry, knowledge management in the field of non-performing asset management is also unique compared to knowledge management in traditional industries. Traditional knowledge management methods and systems cannot be well adapted to the non-performing asset management industry
It is mainly reflected in the following aspects: First, as the object of knowledge management, the knowledge structure of non-performing assets management is relatively complicated
Non-performing assets are usually packaged and sold by banks to asset management companies for disposal. As an asset management company whose main business is the disposal of non-performing assets, there are various ways to dispose of non-performing assets, such as restructuring, debt Therefore, the transaction structure of non-performing asset disposal is diversified, and the knowledge related to non-performing asset disposal covers a wide range and is complex; secondly, the data structure related to knowledge in the field of non-performing asset management is low
Due to the variety of disposal methods, and the customers are mainly medium and large enterprises, the number of customers is relatively small, and the scale of structured data such as transaction record data is small, while due diligence reports, overdue reports, management methods, meeting minutes, etc. , contracts, and comfort letters are all unstructured text data, and the scale is relatively large; third, the knowledge in the field of non-performing asset management is specific and industry-specific
Non-performing asset management business has high entry barriers and low degree of market participation. Therefore, the proportion of non-performing asset management knowledge in the financial field is relatively small, and it is not easy to obtain from the outside world. It relies more on the experience accumulated by the company's practitioners. Therefore, Common knowledge management tools in the financial field, such as industry semantic analysis solutions, industry knowledge organization and management solutions, are not fully applicable to asset management companies' knowledge management of non-performing assets

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A knowledge management method for non-performing assets management based on natural language processing technology
  • A knowledge management method for non-performing assets management based on natural language processing technology
  • A knowledge management method for non-performing assets management based on natural language processing technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0062] In order to illustrate the validity of the patent of the present invention, we verify it based on the text data of the proprietary management method of China Huarong Asset Management Co., Ltd. See Annex 2.

[0063] 1. Data import

[0064] The text data contains a total of 24M texts, 34 data sets, and the formats include word, excel, and pdf; among them, the file in the format of word has a total of 28 data, and the content includes various business work procedures and management methods of each information system. The code is gbk, gb18030; excel has 3 data in total, including risk scoring standards and test sheets; pdf file has 3 data in total, including Huarong's business profile and operating methods.

[0065]Since the excel format files have a table structure and less content, they are manually cleaned; one of the pdf format files is in image format, and the read characters are empty, so it is cleaned. For the rest of the text, use Python software to read and save ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

According to the invention, a non-performing asset management method of an asset management company based on natural language processing technology includes two parts: data import in knowledge base construction, keyword extraction based on deep learning and PageRank. Use a specific lexicon and HMM to discover new word models to segment non-performing asset management knowledge, improve the accuracy of text segmentation, and establish a more complete text lexicon; use deep learning methods to train word vectors, which can avoid word vector representation The "curse of dimensionality" phenomenon can fully mine the information of the lexical context, and can obtain the relationship between words; based on the improved PageRank algorithm, the topological matrix of word connections can be obtained according to the word sequence relationship in a specific contract, and the word vector The cosine value between them is used as the connection weight between words, which not only makes full use of the word vector information obtained from the training lexicon, but also mines the relative positional relationship of words in the text, which has a strong theoretical basis.

Description

technical field [0001] The invention relates to a knowledge management method for non-performing assets management based on natural language processing technology, relates to natural language processing technology in the financial field, and specifically relates to a knowledge management method for asset management companies in the field of non-performing assets management. Background technique [0002] As an integral part of the domestic financial market, the non-performing asset management business plays an important role in maintaining the stability of the financial ecosystem and eliminating systemic risks. It is a stabilizer and fire extinguisher for the smooth operation of the domestic economy. The four major asset management companies (AMCs) initiated and established by the Ministry of Finance are the main body of non-performing asset management. Therefore, carrying out knowledge management of non-performing asset management knowledge in AMC companies will improve the l...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06F16/31G06F40/289G06F16/35G06N5/02
CPCG06F2216/03G06N5/022G06F16/31G06F16/35G06F40/289
Inventor 后其林万谊强仵伟强李峻范小芹路世伦
Owner 华融融通(北京)科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products