Study On Reasoning Of Stockcomment Information And Its System Architecture Based On Semantic

Nowadays, the information for judging the trend of stock held is mainly from the Internet, but the volume of the information is too large and there is no corresponding semantic support among them, it makes stockholders hard to make full use of those stock comments information on the Internet. This paper analyzesd the stock comments of experts and predictsed the trend of stock in next day run so as to guide stockholders to operate stock.On the basic of analyzing semantic Web home and aboard, this paper used semantic to predict the stock comments and gave the system architecture. The main research work is as follows:First, ensure the stock comments and preprocess the stock comments. Through the concept of stock comment information, this paper divided the stock comments into three types:the big board stock comment, the block stock comment and individual stock comment. And the structure of the stock comment information was analyzed, and stock comment information was divided into different kinds of keywords to build my own forecast characteristic thesaurus. This paper stored the stock comment information, which could match the thesaurus, into corresponding database. This paper also introduced the data scrubbing method on the basic of ETL process, preprocessed those disorganized stock comment information in metadatabase, through extracting the data from metadatabae to convert the data to the needed data format, and finally stored these data into the corresponding form.Then, this paper brought forward the inference principle and method of stock comment information. In view of the result (rise, fall, position squaring) may received of stock comment information (the big board stock comment, the block stock comment and individual stock comment) established sample space, and built mathematical model. (1)According to the importance of the three kinds of stock comment information in the process of predicting the trend of stock set initial weight, three entropy spaces (individual stock comment, big board stock comment, corresponding industry plate stock comment) were established, and calculate the entropy of the three spaces; modify weight according to the entropy. (2) Import the theory of entropy weight into the model to establish stock comment system based on entropy weight, and finally get entropy weight to modify the weight. At the same time of the import of maximum entropy theory through subjective restraint condition and objective restraint condition this paper restricts the object function established, (3) The corresponding mathematical model was established and the weight of each evaluate beacon was solved, so that it can make the system self-adaption and get the final analysis of reliability of stock comment information.Finally, the architecture of the stock comment information system was given. This paper divided stock comment information into URI layer, XML/XML Schema layer, Ontology (RDF) layer and Agent layer. It designed XML format of information interchange between each layer in the system, and combining XML and XML schema described interactive process between the user and system. And then this paper built different kinds of ontology in the system and gave the description of the ontology. Then, it described the work mechanism of Agent between ontology, and provided examples of interaction between ontology and Agent. And then it described the design of tie-in module from the interactive mechanism of Web Service. It also provided the whole process of the work order of the system component. At last, this paper through an example verified the feasibility of the inference method of obtaining stock comment information on this architecture.