Font Size: a A A

Design And Implementation Of Annual Report Text And Data Analysis And Visualization

Posted on:2020-05-31Degree:MasterType:Thesis
Country:ChinaCandidate:B W SunFull Text:PDF
GTID:2428330596482452Subject:Computer technology
Abstract/Summary:PDF Full Text Request
The annual report of the enterprise consists of a large amount of text and economic data.These texts and economic data have always been the hotspots in our data analysis and text mining.How to extract this information quickly and accurately,and predict the extracted data,the text is analyzed.the key of.In terms of data research,this paper first uses the LSTM model to predict the stock price and its changing trend of the user's concern.In the experiment,the corresponding features such as the opening price,the closing price,etc.and the label are used to realize the prediction task.The result can be observed.The model can be It predicts the stock price trend for a period of time in the future,and the predicted stock price is also close to the real value.Secondly,in order to better capture the future trend of the industry,we carried out the stock price return ranking forecast,because more users are more concerned about the income of each stock than the stock price,but the traditional regression task can not be considered well in the ranking.The location information,in each iteration,can not make better use of location information for location sorting updates,in order to solve this problem,we introduced a method of sorting learning.Combined with the method of sorting learning,we propose a stock price return forecasting model based on sorting learning to achieve a reasonable forecast of stock price return ranking.The experimental results show that there is a significant improvement effect in the stock price ranking forecasting task,and finally in the visualization link,the forecasting Rankings are displayed.In terms of text processing,we first start the text classification experiment,extract the financial indicator description text in the report from the PDF,and use the naive Bayes classifier to implement the pre-words,stop words and other pre-processing.The two-category experiment of corporate profit loss,and the accuracy rate on the test set is 0.832.Secondly,the relationship between the entity and the entity in the text is extracted.The entity recognition task uses the LTP toolkit based on the annotation recognition method,and successfully identifies the entity name,organization name and other entities in the report,and extracts the Chinese relationship based on the dependency relationship.The method extracts the relationship of the text,obtains the relationship triplet,combines the identified entity,filters out the entity relationship triplet that meets the requirement,and finally uses the enterprise relationship map to visually present,and other experimental results and important Economic data is also visualized.Analyze the economic data of the enterprise report and analyze and extract the text information in the enterprise report,and finally visualize the obtained result.This way of combining theory with practice,while conducting academic research,also makes full use of the results of its research.Reference to the visualized enterprise data text information not only has far-reaching significance for investors who care about the company's business conditions,but also has important significance for the future development of the enterprise and the adjustment of production capacity.
Keywords/Search Tags:data mining, text classification, enterprise annual report, learning to rank
PDF Full Text Request
Related items