基于风险短语挖掘的知识聚合模型研究

Research on Knowledge Aggregation Model Based on Risk Phrase Mining

[目的/意义]大数据时代,金融行业面临着海量多源异构信息源所带来的巨大挑战;针对大数据环境下存在的多源异构金融数据,通过对企业风险知识单元进行挖掘和聚合,从而使其有序化地收敛于高效的金融知识服务,这对于投资决策、风险管理和金融监管等金融决策支持过程具有十分重要的意义。[方法/过程]文章从知识挖掘、知识组织和知识服务的相关理论、方法和技术出发,构建了基于风险短语挖掘的知识聚合模型,该模型主要由知识采集模块、知识挖掘模块和知识服务模块等三大模块所组成。[结果/结论]文章利用N-gram算法来挖掘上市公司年报文本中的候选风险短语,并利用基于统计和基于规则的方法来实现候选短语的过滤,形成可复用的风险短语知识库;将短语作为知识聚合的粒度,利用聚类分析、共现分析和知识检索等技术进行了多种形式的知识聚合,从而为决策者提供智能化的金融知识服务。

[Purpose/significance]In the era of big data,the financial industry faces enormous challenges from voluminous,dynamic and multi-source heterogeneous information sources.Aiming at the massive multi-source heterogeneous financial data under the big data environment,it is very important to mine and aggregate the enterprise risk knowledge units,so that it can converge to efficient financial knowledge services in an orderly manner,which is of great significance for financial decision-making support processes such as investment decision-making,risk management and financial supervision.[Method/process]From the theories,methods and technologies of knowledge mining,knowledge organization and knowledge service,this paper constructs a knowledge aggregation model based on risk phrase mining,which is mainly composed of three modules:knowledge collection module,knowledge mining module and knowledge service module.[Result/conclusion]In this paper,the N-Gram algorithm is utilized to mine candidate risk phrases from annual reports of listed enterprises,and the methods based on combination of statistics and rules are used to filter candidate phrases to form reusable risk phrase knowledge base.This paper takes phrase as the granularity of knowledge aggregation and makes use of cluster analysis,co-word analysis and knowledge retrieval to carry out various forms of knowledge aggregation,so as to provide intelligent financial knowledge services for decision-makers.