Design And Implementation Of Stock Mining And Recognition System In Financial News

With the accumulation of data and the development of intelligent algorithms,user-specific recommendations based on news feed have been spotlight in the industry.In the financial field,the recommendation of financial news also requires personalized services.Information news needs to be cleaned and organized before being recommended to users,which involves the recognition of stock entity in financial information.The recognition of stocks entity is the basis of tasks such as financial information classification,keyword extraction and so on.Limited by resources and domain features,the recognition of stock entity in the financial field can't be solved by a single named entity recognition model,and a targeted strategy needs to design to solve it on the software system.Stock entity in financial information is one of the organization names.The system to recognize stock entity based on expending the dictionary of stock entity and entity disambiguation strategy effectively finishes the recognition of stock entity in the financial field.In the design and implementation of the system,in order to improve the recall rate of entity,the module to mine the alias of stock is constructed offline,and alias mining based on mode of stocks' name mode,based on high frequency suffix and suffix of stocks,based on specific context pattern are implemented.Besides that,in order to recognize the ambiguity of stock entity caused by the matching of string,the effective strategies have been designed to solve entity disambiguation,including synonym discovery and disambiguation,combined ambiguity discovery and disambiguation,entity inclusion relationship disambiguation,English words and digital disambiguation.At the same time,the templated design of the matching process ensure the sufficient expansion space for the future development and iteration of the recognition of stock entity.Comparing with the recognition model of stock entity based on conditional random field,the recognition system of stock entity,which is based on the expansion of the stock dictionary and entity disambiguation,achieved a best result.Through the alias mining optimizing the recall rate and the stock entity disambiguation optimizing the precision rate,the recognition system of stock entity provides a flexible software framework while effectively solving the problem of recognition of stock entity in the financial field,and valuable experience and ideas are provided for the solving of named entity recognition tasks in financial or other fields.