KR20210033294A - Automatic manufacturing apparatus for reports, and control method thereof - Google Patents

Abstract

The present invention relates to a method for automatically writing a report capable of automatically writing an industry and market analysis report by conducting industry and market analysis on an analysis target, automatically extracting the key content from the analysis result, and then writing the implications in textual sentences.

Description

Automatic report generation device and its operation method {AUTOMATIC MANUFACTURING APPARATUS FOR REPORTS, AND CONTROL METHOD THEREOF}

The present invention relates to a method for performing industry/market analysis on an analysis object, automatically extracting core contents from the analysis results, and automatically generating an industry/market analysis report in which the implications are written in text form.

Industry and market analysis is an analysis step that must be taken to determine the feasibility of a business. All technology commercialization agents in industry, academia, research institutes, and governments are responsible for making rational decisions in the entire process of technology commercialization. It is necessary to conduct market analysis.

The method of industry/market analysis can be divided into a qualitative method that relies on the qualitative judgment of an expert or consumer, and a quantitative method based on quantitative data. For example, it is being used more heavily.

However, the quantitative analysis methodology and analysis system adopted in the existing technology have focused only on the purpose of helping to facilitate data collection, processing, and analysis.

In other words, the existing technology is at the level of supporting industry and market analysts to more easily collect, process, and analyze data on the subject of analysis, and furthermore, the analysis results are only summarized in the form of tables or figures. There is a limitation.

The present invention was created in view of the above circumstances, and an object to be reached in the present invention is to perform industry/market analysis on an analysis target, automatically extract core contents from the analysis results, and then describe the implications in text form. It is to automatically generate an industrial and market analysis report written in sentences.

An automatic report generation apparatus according to an embodiment of the present invention for achieving the above object includes: a collection unit for collecting original data related to an analysis object in order to generate a report on the analysis object; A classification unit for classifying the original data for each analysis category designated in the report format and matching it as classification data for each analysis category; And an analysis unit configured to generate analysis data for each analysis category by analyzing the classification data according to category characteristics predefined for each analysis category.

Specifically, the automatic report generation device includes: an extraction unit for extracting core contents from the analysis data for each analysis category; And a generator configured to convert the implications of the core content into a text form according to a predefined form-based character generation rule, and generate a report including the analysis data for each analysis category and the implications converted to the text form. Can include.

Specifically, the collection unit changes or expands the name input for the analysis object to a common name in each data storage on the list, based on a list of data storages designated to match the category characteristics for each analysis category. Original data can be collected.

Specifically, when there is shared data, which is classification data shared between two or more analysis categories, the analysis unit designates a master analysis category having ownership for each of the shared data, and processes the shared data only for the master analysis category. You can make it possible to generate data.

Specifically, in the shared data, ownership for data processing may be assigned to any one of the two or more analysis categories based on at least one of a relationship between data and a data size.

Specifically, the shared data is processed into refined data during a data processing time designated between the two or more analysis categories, and the refined data is when the data processing time ends or data processing is completed. On the other hand, it can be updated for analytics categories that do not have ownership.

Specifically, the extraction unit, according to the content selection weight according to the core content extraction rule for each analysis category, the content of the change in the numerical physical quantity identified from the analysis data, the content of the numerical physical quantity compared between categories, and At least one of the content related to a specific term with the highest exposure frequency can be extracted as the core content.

Specifically, the extraction unit, according to the time selection weight according to the core content extraction rule for each analysis category, at a time before and after a specific event, more recent changes than the past, contents about future prospects, and At least one of the related content can be extracted as the core content.

Specifically, when a hierarchical structure including a high-level category encompassing a plurality of categories of the same level is identified from the analysis data, the extracting unit is based on a category selection weight according to the core content extraction rule for each analysis category. Analysis contents, contents of a new higher-level category newly derived from multiple categories of the same level, contents of a lower-level category showing a difference of more than a set value and statistical physical quantity of the upper-level category, and the analysis target among the same-level categories The content of the category with the highest relevance of the same level, the category with the highest or the lowest physical quantity among the categories of the same level, the content of the category with the physical quantity within the set rank or ratio among the categories of the same level, and the same Among the level categories, at least one of the categories with the greatest change in physical quantity can be extracted as the core content.

In order to achieve the above object, a method of operating an automatic report generation apparatus according to an embodiment of the present invention includes: a collection step of collecting original data related to an analysis object in order to generate a report on the analysis object; A classification step of classifying the original data for each analysis category specified in the report format and matching it as classification data for each analysis category; And an analysis step of analyzing the classification data according to category characteristics predefined for each analysis category, and generating analysis data for each analysis category.

Specifically, the method includes: an extraction step of extracting core content from the analysis data for each analysis category; And a generating step of converting the implications of the core content into a text format according to a predefined format-based character generation rule, and generating a report containing the analysis data for each analysis category and the implications converted to the text format. It may contain more.

Specifically, in the collecting step, based on a list of data stores designated to match the category characteristics for each analysis category, the name input as the analysis object is changed or expanded to a common name in each data store on the list. The original data can be collected.

Specifically, in the analysis step, when there is shared data, which is classification data shared between two or more analysis categories, the shared data is processed only for the master analysis category by designating a master analysis category having ownership for each of the shared data. You can make it possible to generate refined data.

Specifically, in the shared data, ownership for data processing may be assigned to any one of the two or more analysis categories based on at least one of a relationship between data and a data size.

Specifically, the shared data is processed into refined data during a data processing time designated between the two or more analysis categories, and the refined data is when the data processing time ends or data processing is completed. On the other hand, it can be updated for analytics categories that do not have ownership.

Specifically, in the extraction step, according to the content selection weight according to the core content extraction rule for each analysis category, the content of the change in the numerical physical quantity identified from the analysis data, the content of the numerical physical quantity being compared between categories, And at least one of the contents related to the specific term with the highest exposure frequency may be extracted as the core contents.

Specifically, in the extraction step, according to the timing selection weight according to the core content extraction rule for each analysis category, the contents of recent changes rather than the past identified from the analysis data, contents of future prospects, and time points before and after a specific event At least one of the content related to can be extracted as the core content.

Specifically, in the extraction step, when a hierarchical structure including a higher level category encompassing a plurality of categories of the same level is confirmed from the analysis data, the higher level is selected according to the category selection weight according to the core content extraction rule for each analysis category. Analysis content, content of a new higher-level category newly derived from multiple categories of the same level, content of a lower-level category showing a difference greater than or equal to the set value and the statistical physical quantity of the upper-level category, and the analysis target among the categories of the same level The content of the category with the highest relevance to the category, the category with the highest or the lowest physical quantity among the categories of the same level, the content of the category with the physical quantity within the set rank or ratio among the categories of the same level, and Among the categories of the same level, at least one of the categories with the greatest change in physical quantity can be extracted as the core content.

Accordingly, in the automatic report generation device and its operation method of the present invention, the entire process of industry and market analysis can be systematically and efficiently performed through an automated process, so that a number of industries and item fields Also, it enables high-speed and large-capacity analysis in a short time, greatly reducing the time and cost required for industrial and market analysis.

In addition, by providing key features and analysis insights that can be derived from the analysis results beyond the limitations of the quantitative system, which was provided only in the form of tables and figures, through explanatory sentences, it enhances the understanding of information users and relies on the level of interpretation of information users. It does not have the effect of providing more valuable implications.

1 is a schematic configuration diagram for explaining an environment for automatically generating a report according to an embodiment of the present invention.
Figure 2 is a configuration diagram for explaining the configuration of the automatic report generating device according to an embodiment of the present invention.
3 to 16 are exemplary views for explaining a form derived as an analysis result according to an embodiment of the present invention.
17 is an exemplary view for explaining the formation of a report according to an embodiment of the present invention.
18 is a flowchart illustrating a method of operating an automatic report generating apparatus according to an embodiment of the present invention.

Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings.

1 shows an environment for automatically generating a report according to an embodiment of the present invention.

As shown in FIG. 1, the environment for automatically generating a report according to an embodiment of the present invention may include an automatic report generation device 100 that automatically generates a report according to an industry/market analysis result.

The automatic report generation device 100 is a device for performing industry/market analysis and automatically generating a report corresponding to the result of the analysis, and is implemented in the form of a server accessible through a wired/wireless communication network, or a computer system ( Example: computer, mobile phone) It can also be implemented in the form of an in-program.

When the automatic report generation device 100 is implemented in the form of a server, it may be implemented in the form of, for example, a web server, a database server, a proxy server, etc. One or more of a variety of software that enables operation on a network may be installed, and through this, it may be implemented as a computerized system. In addition, the network may be an http network, a private line, an intranet, or any other network. In addition, the connection between each component in the system according to an embodiment of the present invention, data is a random hacker or It can be connected to a secure network so that it is not attacked by other third parties.

Meanwhile, in general, the method of industry/market analysis can be divided into a qualitative method that relies on a qualitative judgment of an expert or a consumer, and a quantitative method based on quantitative data. In recent years, the analysis system is being used more heavily.

However, in the case of such a quantitative analysis methodology and analysis system, the focus is only on the purpose of facilitating the collection, processing, and analysis of data, so industry and market analysts collect, process, and analyze data on the subject of analysis. It is a level that supports the task to be performed more easily, and furthermore, the analysis result is limited to the level of summarizing mainly in the form of tables or figures.

Therefore, when a method that can automatically perform not only the process of collecting, processing, and analyzing industry and market data, but also the process of interpretation and output is proposed, it is very effective not only for the analyst who produces the information, but also for the consumer who uses the analyzed information. It is expected to be a means of supporting analysis.

In addition, if the core content can be automatically extracted and the implications can be derived, the high-level information will be available regardless of the individual analysis capability of the analyst or the interpretation ability of the information user, and furthermore, the meaning of the core content will be interpreted to determine the implications. If it is possible to automatically generate an industry/market analysis report written in text form, it will greatly reduce the time and cost required, especially when creating a report by repeatedly performing industry/market analysis on a number of analysis targets. It is expected to be possible.

Accordingly, in an embodiment of the present invention, an industry/market analysis report is automatically generated in which the analysis target is analyzed for industry/market, core content is automatically extracted from the analysis results, and the implications are written in text form. It is intended to propose a new method for this, and hereinafter will be described in more detail with respect to the configuration of the automatic report generating device 100 for realizing this.

2 shows the configuration of the automatic report generating apparatus 100 according to an embodiment of the present invention.

As shown in FIG. 2, the automatic report generation apparatus 100 according to an embodiment of the present invention includes a collection unit 10 for collecting data on analysis data, a classification unit 20 for classifying the collected data, And an analysis unit 30 for analyzing the classified data.

In addition, the automatic report generation apparatus 100 according to an embodiment of the present invention may further include a configuration of an extraction unit 40 for extracting core contents and a generation unit 50 for generating a report in addition to the above-described configuration. have.

The entire configuration or at least a part of the automatic report generation device 100 including the above collection unit 10, classification unit 20, analysis unit 30, extraction unit 40, and generation unit 50 is a hardware module It may be implemented in the form of a form or a software module, or may be implemented in a form in which a hardware module and a software module are combined.

Here, the software module may be understood as, for example, an instruction executed by a processor that controls an operation in the automatic report generating device 100, and this instruction is a form mounted in the memory in the automatic report generating device 100. You will be able to have it.

On the other hand, the automatic report generation apparatus 100 according to an embodiment of the present invention further includes a configuration of the communication unit 60, which is an RF module responsible for a practical communication function with a remote device through a wired or wireless communication network, in addition to the above configuration. can do.

Here, the communication unit 60 includes, but is not limited to, an antenna system, an RF transceiver, one or more amplifiers, a tuner, one or more oscillators, a digital signal processor, a codec chipset, and a memory. Any known circuit to perform may be included.

As described above, the automatic report generation apparatus 100 according to an embodiment of the present invention performs an industry/market analysis on an analysis target, automatically extracts the core contents from the analysis results, and then writes its implications as text-type sentences. The market analysis report can be automatically generated. Hereinafter, each component in the automatic report generation device 100 for realizing this will be described in more detail.

The collection unit 10 performs a function of collecting data for analysis data.

More specifically, the collection unit 10 collects data related to the analysis target (hereinafter, source data) in order to generate a report on the analysis target.

At this time, the analysis target may be input from the user along with information on the report format (composition) as information on the industry or item, and the report format here may be designated to include two or more analysis categories according to user designation. have.

Accordingly, when an analysis target is input, the collection unit 10 checks the list of designated data stores to match the characteristics of each category for each analysis category specified in the report format, and stores the name input as the analysis target on the list. Original data can be collected by changing or expanding to a common name (eg, classification code) in each.

That is, the collection unit 10 collects the original data from the designated data storage matching each analysis category specified in the report format when collecting the original data for the analysis target, thereby collecting additional data other than the original data suitable for the analysis category. By limiting the data, the time required to collect and analyze the original data can be minimized.

For example,'elevator' is input as an analysis target, and as the format of the report, Chapter 1 Definition and Overview, Chapter 2 Environmental Analysis, Chapter 3 Market Structure Analysis, Chapter 4 Competition Status Analysis, Chapter 5 Markets Analysis categories can be specified in the order of size estimation and forecast, Chapter 6 Financial Structure Analysis, and Chapter 7 Implication Index Analysis.

In this case, the data repositories of the Korea Institute of Science and Technology Information, KMAPS, the National Statistical Office, and news articles may be matched to the analysis category corresponding to the definition and outline of Chapter 1. KMAPS of Korea Institute of Technology and Information, news articles, internet encyclopedias, and data repositories of government departments and public institutions can be matched.

In addition, the data storage of the Korea Institute of Science and Technology Information KMAPS and the Bank of Korea can be matched to the classification category corresponding to Chapter 3 Market Structure Analysis, and in the classification category corresponding to Chapter 4 Competition Status Analysis, Korea Science and Technology Information Researcher KMAPS, Financial Supervisory Service, and private credit rating agencies' data repositories can be matched.

Furthermore, the data storage of the Korea Institute of Science and Technology Information KMAPS, the National Statistical Office, the Customs Service, and the foreign national statistical offices including the United States can be matched to the classification category corresponding to Chapter 5 Market Size Estimation and Prospect. A private credit rating agency may be matched to a corresponding classification category, and a data repository of KMAPS of the Korea Institute of Science and Technology Information may be matched to a classification category corresponding to Chapter 7 Analysis of Implications Indicators.

On the other hand, if'elevator' is entered as an analysis target, it is a classification code that corresponds to the classification target. (Lift and Elevator Installation Construction Service), 87158 (Elevator and Escalator Maintenance and Repair Service), North American Standard Industry Classification 333921 (Elevator and Moving Stairway Manufacturing) can be selected.

Based on this, the market size data classified and organized in advance for the Korean standard industry classification'elevator manufacturing industry' from the KMAPS (KISTI Market Analysis and Prediction System) webpage, the industry and market analysis system of the Korea Institute of Science and Technology Information, and sales by company. You can select and download structured data such as data and inter-industry transaction data, and from the web page of the National Statistical Office, you can web crawl unstructured data for industry descriptions such as the definition, specific scope, and major products of the industry. For lower-level items, you can select and download structured data on the amount of shipment, the number of companies, and the number of workers classified in advance.

In addition, you can select and download structured data on major financial items and industry-related analysis classified in advance for the industry from the web page of the Bank of Korea, and the industry is classified in advance from the web page of the Financial Supervisory Service. You can select and download structured data and unstructured data on the outline and finance of the engaged companies.

In addition, you can select and download structured data about the financial and the outline of the companies classified in advance for the industry from the webpage of private credit rating agencies, and the North American Standard Industry Classification 333921 can be downloaded from the web page of the U.S. Statistical Office. You can select and download structured data on shipment amount, number of companies, number of workers, import and export, etc. classified and organized in the Korea International Trade Association's webpage, by selecting the import and export data classified in advance as MTI classification 745100 to 745200 from the webpage of the Korea International Trade Association. You can download it.

Furthermore, major news articles and unstructured data on definitions and characteristics derived from search terms such as'lift','elevator', and'escalator' related to the analysis target'elevator' are retrieved from the news portal and internet encyclopedia webpage. Web crawling is possible, and unstructured data derived by searching for the above related keywords from information sources for industry and market analysis reports published by government ministries and public institutions, such as the Small and Medium Business Technology Roadmap Report of the Ministry of SMEs and Startups, can be downloaded or web crawled.

The classification unit 20 performs a function of matching classification data for each analysis category.

More specifically, when the collection of original data for each analysis category specified in the report format is completed, the classification unit 20 classifies each original data and matches it as classification data for each analysis category.

As in the previous example, the'lift' is input as an analysis target, and as the format of the report, Chapter 1 Definition and Overview, Chapter 2 Environmental Analysis, Chapter 3 Market Structure Analysis, Chapter 4 Competition Status Analysis, and Chapter 5 If the analysis category is specified in the order of Chapter 1, the estimation and forecast of market size, Chapter 6 Financial Structure Analysis, and Chapter 7 Implication Index Analysis, the analysis categories corresponding to Chapter 1 Definition and Overview include industry explanatory data from the National Statistical Office, and news articles. Data can be matched, and in the analysis category corresponding to Chapter 2 Environmental Analysis, news article data, Internet encyclopedia data, and industry/market analysis reports published by government departments and public institutions can be classified and matched.

In addition, in the classification category corresponding to Chapter 3 Market Structural Analysis, the inter-industry transaction data of KMAPS of Korea Institute of Science and Technology Information and the data of industry-related analysis of the Bank of Korea can be matched, and the classification corresponding to Chapter 4 competition status analysis. In the category, sales data for each company of KMAPS of the Korea Institute of Science and Technology Information, corporate financial data of the Financial Supervisory Service, and corporate financial data of a private credit rating company may be matched.

Furthermore, in the classification category corresponding to Chapter 5 Market Size Estimation and Prospect, the market size data of KMAPS of the Korea Institute of Science and Technology Information, shipment amount data of the National Statistical Office, import and export data of the Korea Customs Service, and the shipment amount and import and export data of the US Statistical Office can be matched. The classification category corresponding to Chapter 6 Financial Structure Analysis can be matched with corporate financial data of private credit rating agencies, and the KMAPS Index data of KMAPS of Korea Institute of Science and Technology Information can be matched with the classification category corresponding to Chapter 7 Implication Index Analysis.

The analysis unit 30 performs a function of generating analysis data for each analysis category.

More specifically, when the analysis data classification for each analysis category specified in the report format is completed, the analysis unit 30 analyzes the classification data classified according to the characteristics of each category to generate analysis data for each analysis category.

In this case, in generating the analysis data for each analysis category, the analysis unit 30 may perform a data processing procedure to refine the analysis data into refined data in a form suitable for data analysis for each analysis category.

For example, in the case of numerical data on market size or major financial items, the unit is unified in billions, and in relation to the processing of missing values, in the case of continuous data, average substitution, regression substitution, most likelihood substitution, and multiple substitution are used. In the case of categorical data, it replaces missing values through mode substitution, supervised learning, and semi-supervised learning, and cases with values outside a predetermined range in relation to outlier processing, or a specific percentage of all cases. An exception can be handled by detecting a case with a value within the range or a case identified as an outlier as a result of statistical analysis.

Also, between the categories of analysis, Chapter 1. For market size and forecast, heterogeneous data such as market size data of KMAPS of Korea Institute of Science and Technology Information, shipment amount data of Statistics Korea, import and export data of Korea Customs Service, shipment amount and import and export data of US Statistical Office are matched, linked, and calculated according to predetermined rules. can do.

In this regard, [Table 1] shows the results of collecting the data on shipments classified and organized into the Korean Standard Industry Classification'Elevator Manufacturing Industry' from the National Statistical Office's web page, and [Table 2] shows the MTI classification 745100 through the web page of the Korea International Trade Association. It shows the result of collecting the imported and exported data classified as 745200.

Korean Standard Industry Classification “Elevator Manufacturing” Shipment Scale (Unit: KRW million) year 2007 2008 2009 2010 2011 2012 2013 Shipment amount 2,795,480 2,790,190 2,682,510 2,889,870 3,214,950 3,334,220 2,722,460 year 2014 2015 2016 2017 Shipment amount 3,300,340 3,789,260 4,238,910 4,828,500

Scale of import and export of MTI classification 745100 to 745200 (Unit: thousand dollars) year 2007 2008 2009 2010 2011 2012 2013 Exports 745 100 191,111 235,005 239,516 70,669 74,944 69,323 72,115 Exports 745200 5,382 6,915 824 2,842 284 2,857 699 Income 745 100 32,349 11,669 13,553 14,241 12,548 5,045 10,615 Income 745200 39,249 51,386 42,427 46,432 54,989 38,750 41,861 year 2014 2015 2016 2017 2018 Exports 745 100 54,290 86,514 56,408 47,819 52,824 Exports 745200 0 0 36 90 0 Income 745 100 12,521 6,003 6,753 1,656 1,224 Income 745200 49,692 75,753 61,350 53,667 51,617

Shipment amount by detailed item related to the Korean Standard Industry Classification'elevator manufacturing industry' (unit: KRW million) year 2007 2008 2009 2010 2011 2012 2013 elevator 12,624 14,089 12,129 12,277 13,208 8,791 11,525 escalator 1,117 236 2,055 1,651 1,818 1,951 1,224 Lift equipment 1,346 1,525 1,227 703 1,045 659 510 Parking machine 873 686 1,205 1,946 1,245 7,096 1,401 Elevator parts 3,382 3,766 3,464 4,276 4,220 5,279 3,969 year 2014 2015 2016 2017 elevator 14,660 20,818 21,363 24,333 escalator 902 - 822 - Lift equipment 425 969 1,211 1,488 Parking machine 1,150 1,771 2,307 2,522 Elevator parts 4,975 5,272 6,073 7,658

As shown in [Table 1], we selected and collected the shipment amount data collected in units of million won for the Korean Standard Industry Classification'elevator manufacturing industry' from the National Statistical Office. In addition, as shown in [Table 3], the'elevator manufacturing industry' and As shown in [Table 2], the shipment amount data for related detailed items is selected and collected, and as shown in [Table 2], the amount of exports and imports collected in units of thousands of dollars for MTI classifications 745100 to 745200 from the Korea International Trade Association is selected and collected. ], the estimated shipment value of escalator items in 2015 and 2017 was replaced by KRW 900 million and KRW 800 million, respectively, through moving average substitution, and the outlier detection modeling confirmed that no outliers existed, and between shipments and trade amounts After linking the data, selecting and collecting annual average won/dollar exchange rate data from the Bank of Korea, unifying the monetary unit between the shipment amount and the trade amount into billion won, then processing it through an operation process that subtracts the export amount from the shipment amount and adds the import amount. As the resulted data, the size of the domestic market of the'lift', which is the target of analysis, can be newly produced as shown in [Table 4].

'Elevator' Domestic Market Scale (Unit: KRW 100 million) year 2007 2008 2009 2010 2011 2012 2013 Shipment amount 26,794 25,930 24,472 28,750 32,064 33,022 27,002 year 2014 2015 2016 2017 Shipment amount 33,087 37,839 42,524 48,369

In addition, when generating analysis data for each analysis category, the analysis unit 30 may perform natural language processing on unstructured data.

For example, morpheme analysis, term extraction, term frequency analysis, term-document matrix generation, simultaneous occurrence frequency analysis between terms, for unstructured data on the company's major product names among corporate data collected for competition status analysis. Natural language processing such as relationship analysis, similarity analysis between documents, and topic modeling can be performed.

In addition, in generating the analysis data for each analysis category, the analysis unit 30 may diagram the analysis result in the form of a table or picture according to the characteristics of each analysis category.

For example, Chapter 5. For analysis categories corresponding to market size and forecast, time series analysis such as trend method, averaging method, smoothing method, and cumulative autoregressive moving average (ARIMA) based on the domestic market size estimated by calculating the shipment amount and export/import amount of the analysis target. As shown in Figs. 3 and 4, the future market size is predicted using a method and a method using diffusion models such as BASS, Gomperz, and Logistics, and the predicted result by the cumulative autoregressive moving average method, which is the most suitable method. Likewise, it can be plotted in the form of tables and figures, and the analysis results such as dependence on imports, export share, and trade-specific index, which can be additionally calculated through calculation of shipments and imports and exports, are plotted in the form of tables and figures. can do.

Also, Chapter 3. For the analysis category corresponding to the market structure analysis, based on the refined data on the Korean standard industry classification'elevator manufacturing industry', the downstream industry that supplies parts to the industry and the downstream industry that consumes elevators produced in the same industry are analyzed. By performing the analysis of the front and rear industrial structure, it can be diagrammed in the form of a picture as shown in FIG. 5.

Also, Chapter 4. For the analysis category corresponding to the competition status analysis, based on refined and processed data, the market share based on sales of companies participating in the Korean Standard Industry Classification'elevator manufacturing industry' is estimated and plotted as shown in Fig.6. Based on the results of the market share distribution by company, the market concentration was analyzed through indicators such as HHI (Hirshman-Herfindahl Index) and CR (Concentration Ratio)3 newly calculated and generated by the processing department, as shown in FIG. It can be illustrated as a picture, and the market share by company size divided into large, mid-sized, mid-sized, and small-sized companies, which are higher-level categories than the corporate-level category, can be analyzed and plotted as a picture as shown in FIG. As shown in Fig. 9, it is possible to compare and analyze major financial items of companies in the last three years and plot them in the form of a table, and the data processed through natural language processing of unstructured data on major product names of companies participating in the same industry is used as a word. It can be analyzed in the cloud and plotted in the form of a picture as shown in FIG. 10, and the number of employees and the distribution of business power of companies participating in the same industry can be analyzed and plotted in the form of a picture as shown in FIG.

Also, Chapter 6. For the analysis category corresponding to the financial structure analysis, based on refined data, the average major of all companies participating in the Korean Standard Industrial Classification'elevator manufacturing industry', companies within the top 25% of operating profits, SMEs, and companies with less than 5 years of founding. The financial ratio can be analyzed and plotted in the form of a table as shown in FIG. 12, and the major financial ratios can be analyzed in a time series and plotted in the form of a figure as shown in FIG.

Further, Chapter 7. As for the analysis category corresponding to the analysis of the implication index, as shown in Fig. 14, by synthesizing the analysis contents of Chapters 2 to 6, analyzing the new entry status taking into account the share of sales and the number of companies within 3 years of the founding of the company. It can be plotted as a figure, and it can be plotted as a figure as shown in Fig. 15 by analyzing the growth opportunity that takes into account market concentration, market growth rate, and participation in SMEs, and profits taking into account the estimated market size and operating margin of sales after 5 years. By analyzing the possibility, it can be schematically illustrated as shown in FIG. 16.

Meanwhile, in the case of analysis data classified by analysis category according to an embodiment of the present invention, it is considered that shared data, which is analysis data shared with each other, exists between analysis categories due to the similarity of characteristics between each category.

Accordingly, when there is shared data, which is classification data shared between two or more analysis categories, the analysis unit 30 designates a master analysis category that has ownership for each shared data, and processes the shared data only for the designated master analysis category. To be able to create.

That is, in an embodiment of the present invention, by granting ownership, which is the authority for data purification, to only one of the analysis categories that share shared data with each other, the data is better than when data purification is performed on the shared data for each analysis category. It is possible to significantly reduce the use of resources required for refining.

In this case, the shared data may be assigned ownership for data processing to any one of two or more analysis categories based on at least one of, for example, a relationship between data and a data size, and after the ownership assignment, between the two or more analysis categories Data processing into refined data may be performed during a specified data processing time, and when data processing time ends or data processing is completed, relative analysis categories that do not have ownership may be automatically updated.

That is, in an embodiment of the present invention, when data purification is completed in each analysis category having ownership of the shared data, the purification data is shared with another analysis category that shares the shared data before purification, so that the other analysis category also It can support to derive analysis results from refined data.

The extraction unit 40 performs a function of extracting core contents.

More specifically, when the generation of the analysis data, which is the analysis result of the classification data for each analysis category, is completed, the extraction unit 40 extracts core contents from the analysis data of each analysis category.

At this time, when the hierarchical structure including a higher level category encompassing a plurality of categories of the same level is identified from the analysis data, the extraction unit 40 determines the higher level according to the category selection weight according to the core content extraction rule for each analysis category. Analysis contents, contents of a new higher-level category newly derived from multiple categories of the same level, contents of a lower-level category showing a difference of more than a set value and statistical physical quantity of the upper-level category, and the analysis target among the same-level categories The content of the category with the highest relevance of the same level, the category with the highest or the lowest physical quantity among the categories of the same level, the content of the category with the physical quantity within the set rank or ratio among the categories of the same level, and the same Among the level categories, at least one of the categories with the greatest change in physical quantity can be extracted as the core content.

In other words, in terms of content selection, the content of the physical quantity and its change expressed in numerical values such as market size, market concentration, sales, and asset size, and the contents of comparing the physical quantity and its change between specific categories, most often appear. The content related to the term can be extracted as the core content, and in terms of time selection, content about recent changes rather than the past, content about future prospects, content about the time when there was a sudden change in physical quantity, before and after a specific event. The content of a point in time can be extracted as the core content, and in terms of selection of categories and levels, the high-level that can explain the overall trend of the analysis result that includes the contents of the upper level and the lower level belonging to it at the same time. In the case where the analysis contents of the analysis and the higher-level categories that will include multiple categories of the same level are not previously determined, new higher-level categories are newly established through unsupervised learning modeling such as K-means clustering, two-stage clustering, and Kohonen clustering. Contents corresponding to the upper-level categories configured and configured, the lower-level categories with statistically significant differences from the statistical values of the physical quantities of the upper-level categories, and the categories most relevant to the analysis target among the same-level categories, the same Among the level categories, the contents of the category with the highest or lowest physical quantity, the physical quantity within a certain rank or ratio, and the category with the most rapid change among the same level categories can be extracted as core contents.

For reference, in terms of selection of the extraction algorithm, statistical techniques such as variance analysis, correlation analysis, regression analysis, and discriminant analysis are used, or by applying machine learning algorithms such as supervised learning, semi-supervised learning, and unsupervised learning, the core content extraction rule is created. can do.

For example, in one embodiment of the present invention, numerical data on the recent domestic market size, growth rate, and future forecast for the Korean standard industry classification'elevator manufacturing industry' can be extracted as core contents. The contents of the term'elevator', which is the most common term among items, and the contents of comparison with the contents of the same level category,'escalator', can be extracted as the core contents.

In addition, it is possible to extract the contents of the change in total assets over the last three years from the financial information of the companies with total assets within the 10th place among participating companies in the same industry, and is the most rapid among the average total asset turnover time series analysis data of the participating companies in the same industry. The content of the 2014 data showing an increase can be selected as the core content.

In addition, the contents of the change in the average sales growth rate of the same industry at the time before and after the global financial crisis in 2009 can be extracted as the core contents, and the'elevator manufacturing industry', which shows the overall trend among the contents of the'elevator manufacturing industry' and its sub-related detailed products. 'Shipment amount data can be selected as the core content.

In addition, as a result of clustering the floating ratio time series data patterns of all industries in the manufacturing industry through K-means clustering modeling, the characteristics of the C-type pattern to which the same industry belongs can be selected as the core content. The core contents of the financial analysis of Company C, whose profit margin is statistically significantly low, can be selected.

In addition, you can select as the core content the analysis of the bargaining power of the'apartment construction industry', which has the largest transaction volume among the demand industries of the industry, and the'primary metal product manufacturing industry,' which has the largest transaction volume among the parts supply industries. Among the various indicators for the average financial ratio of Korea, you can select the contents of the sales growth rate, which showed the most recent change, as the core content.

The generation unit 50 performs a function of automatically generating a report.

More specifically, when the extraction of the core content for each analysis category is completed, the generation unit 50 converts the implications of the core content into a text format according to a predefined form-based character generation rule, and analyzes data for each analysis category. And a report containing the implications converted to the above text format is generated.

For example, in one embodiment of the present invention, for the part that predicts the market size and growth rate of the next five years extracted as the core content in Chapter 5 Market Size and Prospect, “[The name of the industry to be analyzed (industry code)] The domestic market size of [Analysis Base Year] is the level of [Analysis Base Year's Domestic Market Size], and is the annual average growth rate of [Expected annual growth rate for the next four years] [(Optional) Range A: greatly increased, Range B: increased, C range: slightly increased, D range: maintained, E range: slightly decreased, F range: decreased, G range: significantly decreased], expected to reach the level of [four years from the base year] [4 years later domestic market size] Using a pre-determined format-based sentence generation rule, such as “It will be”, the domestic market size of “the elevator manufacturing industry (C29162) is 48,369 billion won in 2017, and it has increased at an annual average growth rate of 4.2% to reach 5699 billion won in 2021. It is expected that this is expected.” can be automatically generated.

Also in Chapter 6. It is the content of the recent change trend of the average financial ratio extracted as a core content in the financial structure analysis. Looking at the time series data of the major financial ratios for the last 10 years for “[Analysis target industry name (industry code)], recently [( Select) A: Sales growth rate, B: Operating margin, C: Current ratio, D: Total asset turnover] is [(Optional) A range: rapidly increasing, B range: increasing, C range: slightly increasing, D range: {last 3 Yearly average rate} is maintained, E range: slightly decrease, F range: decrease, G range: rapidly decrease] by using a pre-determined format-based sentence generation rule, Looking at the time series data of the financial ratio over the past 10 years, in recent years, the rate of increase in sales is rapidly decreasing, the operating margin is increasing rapidly, the current ratio is increasing rapidly, and the total asset turnover rate is increasing somewhat” automatically. Can be generated.

In addition, natural language processing of sentences contained in industry and market analysis reports, and deep learning in units of words or syllables, complements the above style-based sentence generation rules by concatenating the words or syllables most likely to appear after a specific word or syllable. can do.

On the other hand, in an embodiment of the present invention, each position can be determined by combining the analyzed table or picture and text in the form of text, and a report can be generated. By arranging explanatory sentences in which the contents are converted into text in a predetermined order, a report may be generated in the form as shown in FIG. 17, for example.

As described above, according to the configuration of the automatic report generation device 100 according to an embodiment of the present invention, after collecting, classifying, refining, and processing quantitative data on an analysis target, industry and market analysis is automatically performed. And, by automatically extracting the core content from the analysis results and automatically generating an industry/market analysis report that interprets the meaning and writes the implications in text form, it is very useful not only for analysts who produce information, but also for consumers who use the analyzed information. It can be an effective means of supporting industry and market analysis. In addition, by automatically extracting the core content and deriving implications, high-level information can be utilized regardless of the difference in the analyst's personal analysis capability or the information user's interpretation ability. In addition, by interpreting the meaning of the core content, it automatically generates an industry/market analysis report with implications in text form, especially when creating a report by repeatedly performing industry/market analysis on a number of analysis targets. It can greatly reduce the time and cost of becoming.

Hereinafter, a description of the operation method of the automatic report generation apparatus 20 according to an embodiment of the present invention will be continued with reference to FIG. 18.

First, the collection unit 10 collects original data related to the analysis target to generate a report on the analysis target (S10).

At this time, the analysis target may be input from the user along with information on the report format (composition) as information on the industry or item, and the report format here may be designated to include two or more analysis categories according to user designation. have.

Accordingly, when an analysis target is input, the collection unit 10 checks the list of designated data stores to match the characteristics of each category for each analysis category specified in the report format, and stores the name input as the analysis target on the list. Original data can be collected by changing or expanding to a common name (eg, classification code) in each.

That is, the collection unit 10 collects the original data from the designated data storage matching each analysis category specified in the report format when collecting the original data for the analysis target, thereby collecting additional data other than the original data suitable for the analysis category. By limiting the data, the time required to collect and analyze the original data can be minimized.

Then, when the collection of the original data for each analysis category specified in the report format is completed, the classification unit 20 classifies each original data and matches it as classification data for each analysis category (S20).

Subsequently, when the analysis data classification for each analysis category specified in the report format is completed, the analysis unit 30 analyzes the classification data classified according to the characteristics of each category, and generates analysis data for each analysis category (S30).

In this case, in generating the analysis data for each analysis category, the analysis unit 30 may perform a data processing procedure to refine the analysis data into refined data in a form suitable for data analysis for each analysis category.

For example, in the case of numerical data on market size or major financial items, the unit is unified in billions, and in relation to the processing of missing values, in the case of continuous data, average substitution, regression substitution, most likelihood substitution, and multiple substitution are used. In the case of categorical data, it replaces missing values through mode substitution, supervised learning, and semi-supervised learning, and cases with values outside a predetermined range in relation to outlier processing, or a specific percentage of all cases. An exception can be handled by detecting a case with a value within the range or a case identified as an outlier as a result of statistical analysis.

Meanwhile, in the case of analysis data classified by analysis category according to an embodiment of the present invention, it is considered that shared data, which is analysis data shared with each other, exists between analysis categories due to the similarity of characteristics between each category.

Accordingly, when there is shared data, which is classification data shared between two or more analysis categories, the analysis unit 30 designates a master analysis category that has ownership for each shared data, and processes the shared data only for the designated master analysis category. To be able to create.

That is, in an embodiment of the present invention, by granting ownership, which is the authority for data purification, to only one of the analysis categories that share shared data with each other, the data is better than when data purification is performed on the shared data for each analysis category. It is possible to significantly reduce the use of resources required for refining.

In this case, the shared data may be assigned ownership for data processing to any one of two or more analysis categories based on at least one of, for example, a relationship between data and a data size, and after the ownership assignment, between the two or more analysis categories Data processing into refined data may be performed during a specified data processing time, and when data processing time ends or data processing is completed, relative analysis categories that do not have ownership may be automatically updated.

That is, in an embodiment of the present invention, when data purification is completed in each analysis category having ownership of the shared data, the purification data is shared with another analysis category that shares the shared data before purification, so that the other analysis category also It can support to derive analysis results from refined data.

In addition, when generating analysis data for each analysis category, the analysis unit 30 may perform natural language processing on unstructured data.

For example, morpheme analysis, term extraction, term frequency analysis, term-document matrix generation, simultaneous occurrence frequency analysis between terms, for unstructured data on the company's major product names among corporate data collected for competition status analysis. Natural language processing such as relationship analysis, similarity analysis between documents, and topic modeling can be performed.

In addition, in generating the analysis data for each analysis category, the analysis unit 30 may diagram the analysis result in the form of a table or picture according to the characteristics of each analysis category.

Further, when the generation of the analysis data, which is the analysis result of the classification data for each analysis category, is completed, the extraction unit 40 extracts core contents from the analysis data of each analysis category (S40).

At this time, when the hierarchical structure including a higher level category encompassing a plurality of categories of the same level is identified from the analysis data, the extraction unit 40 determines the higher level according to the category selection weight according to the core content extraction rule for each analysis category. Analysis contents, contents of a new higher-level category newly derived from multiple categories of the same level, contents of a lower-level category showing a difference of more than a set value and statistical physical quantity of the upper-level category, and the analysis target among the same-level categories The content of the category with the highest relevance of the same level, the category with the highest or the lowest physical quantity among the categories of the same level, the content of the category with the physical quantity within the set rank or ratio among the categories of the same level, and the same Among the level categories, at least one of the categories with the greatest change in physical quantity can be extracted as the core content.

In other words, in terms of content selection, the content of the physical quantity and its change expressed in numerical values such as market size, market concentration, sales, and asset size, and the contents of comparing the physical quantity and its change between specific categories, most often appear. The content related to the term can be extracted as the core content, and in terms of time selection, content about recent changes rather than the past, content about future prospects, content about the time when there was a sudden change in physical quantity, before and after a specific event. The content of a point in time can be extracted as the core content, and in terms of selection of categories and levels, the high-level that can explain the overall trend of the analysis result that includes the contents of the upper level and the lower level belonging to it at the same time. In the case where the analysis contents of the analysis and the higher-level categories that will include multiple categories of the same level are not previously determined, new higher-level categories are newly established through unsupervised learning modeling such as K-means clustering, two-stage clustering, and Kohonen clustering. Contents corresponding to the upper-level categories configured and configured, the lower-level categories with statistically significant differences from the statistical values of the physical quantities of the upper-level categories, and the categories most relevant to the analysis target among the same-level categories, the same Among the level categories, the contents of the category with the highest or lowest physical quantity, the physical quantity within a certain rank or ratio, and the category with the most rapid change among the same level categories can be extracted as core contents.

For reference, in terms of selection of the extraction algorithm, statistical techniques such as variance analysis, correlation analysis, regression analysis, and discriminant analysis are used, or by applying machine learning algorithms such as supervised learning, semi-supervised learning, and unsupervised learning, the core content extraction rule is created. can do.

Thereafter, when the extraction of the core content for each analysis category is completed, the generation unit 50 converts the implications of the core content into a text form according to a predefined form-based character generation rule, and the analysis data for each analysis category and the above A report stating the implications converted into text form is generated (S50-S60).

As described above, according to the operation method of the automatic report generation device 100 according to an embodiment of the present invention, after collecting, classifying, refining, and processing quantitative data on an analysis target, the industry and market analysis are automatically performed. After performing the analysis and automatically extracting the core content from the analysis results, by automatically generating an industry/market analysis report that interprets the meaning and writes the implications in text form, not only the analyst who produces the information, but also the consumers who use the analyzed information. It can be a very effective means of supporting industry and market analysis. In addition, by automatically extracting the core content and deriving implications, high-level information can be utilized regardless of the difference in the analyst's personal analysis capability or the information user's interpretation ability. In addition, by interpreting the meaning of the core content, it automatically generates an industry/market analysis report with implications in text form, especially when creating a report by repeatedly performing industry/market analysis on a number of analysis targets. It can greatly reduce the time and cost of becoming.

Meanwhile, the steps of the method or algorithm described in connection with the embodiments presented herein may be directly implemented in hardware or implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded in the medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those produced by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

Until now, the present invention has been described in detail with reference to preferred embodiments, but the present invention is not limited to the above-described embodiments, and the technical field to which the present invention belongs without departing from the gist of the present invention claimed in the following claims. Anyone of ordinary skill in the art will say that the technical idea of the present invention extends to the range in which various modifications or modifications are possible.

According to the automatic report generation apparatus and its operation method according to the present invention, an industry/market analysis is performed on an analysis target, and core contents are automatically extracted from the analysis results, and the implications are written in text form for an industry/market analysis. In the sense that reports can be automatically generated, as it goes beyond the limitations of existing technologies, the possibility of marketing or sales of applied devices is sufficient as well as the extent to which it can be carried out clearly in reality, as well as the use of related technologies. It is an invention that can be used in the future.

100: automatic report generation device
10: collection unit 20: classification unit
30: analysis unit 40: extraction unit
50: generation unit

Claims (18)

A collection unit that collects original data related to the analysis target to generate a report on the analysis target;
A classification unit for classifying the original data for each analysis category designated in the report format and matching it as classification data for each analysis category; And
And an analysis unit configured to generate analysis data for each analysis category by analyzing the classification data according to category characteristics predefined for each analysis category. The method of claim 1,
The automatic report generation device,
An extraction unit for extracting core contents from the analysis data for each analysis category; And
Further comprising a generating unit for converting the implications of the core content into a text format according to a predefined format-based character generation rule, and generating a report containing the analysis data for each analysis category and the implications converted to the text format. Automatic report generation device, characterized in that. The method of claim 1,
The collection unit,
The original data is collected by changing or expanding the name input for the analysis object to a common name in each data storage on the list, based on a list of data storages designated to match the category characteristics for each analysis category. Automatic report generation device. The method of claim 1,
The analysis unit,
When there is shared data, which is classification data shared between two or more analysis categories, a master analysis category having ownership for each shared data is designated so that refined data processed by processing the shared data can be generated only for the master analysis category. Automatic report generation device, characterized in that. The method of claim 4,
The shared data,
An automatic report generation apparatus, characterized in that ownership for data processing is allocated to any one of the two or more analysis categories based on at least one of a degree of relationship between data and a data size. The method of claim 4,
The shared data,
Data processing into refined data is performed during the data processing time specified between the two or more analysis categories,
The refinement data,
When the data processing time ends or the data processing is completed, the automatic report generation apparatus is updated for an analysis category that does not have ownership. The method of claim 2,
The extraction unit,
According to the content selection weight according to the core content extraction rule for each analysis category, the content of the change in the numerical physical quantity identified from the analysis data, the content of the numerical physical quantity compared between categories, and the specific term with the highest exposure frequency Automatic report generation device, characterized in that extracting at least one of the content related to the core content. The method of claim 2,
The extraction unit,
According to the time selection weight according to the core content extraction rules for each analysis category, at least one of the contents of recent changes rather than the past identified from the analysis data, contents of future prospects, and contents of the time before and after a specific event is the core. Automatic report generation device, characterized in that extracting as content. The method of claim 2,
The extraction unit,
When a hierarchical structure including a high-level category encompassing multiple categories of the same level is identified from the analysis data, the analysis content for the high level, multiple of the same level, according to the category selection weight according to the core content extraction rule for each analysis category. The content of the new upper-level category newly derived from the category, the content of the lower-level category showing a difference greater than or equal to the set value and the statistical physical quantity of the upper-level category, the category with the highest relevance to the analysis object among the categories of the same level The content of the category with the highest or lowest physical quantity among the categories of the same level, the content of the category with the physical quantity within the set rank or ratio among the categories of the same level, and the change of the physical quantity among the categories of the same level Automatic report generation device, characterized in that extracting at least one of the contents of the largest category as the core contents. A collection step of collecting original data related to the analysis target to generate a report on the analysis target;
A classification step of classifying the original data for each analysis category specified in the report format and matching it as classification data for each analysis category; And
And an analysis step of analyzing the classification data according to the category characteristics predefined for each analysis category, and generating analysis data for each analysis category. The method of claim 10,
The above method,
An extraction step of extracting core contents from the analysis data for each analysis category; And
Further, a generating step of generating a report containing the analysis data for each analysis category and the implications converted to the text format by converting the implications of the core content into a text format according to a predefined format-based character generation rule. Operating method of the automatic report generating device comprising a. The method of claim 10,
The collecting step,
The original data is collected by changing or expanding the name input for the analysis object to a common name in each data storage on the list, based on a list of data storages designated to match the category characteristics for each analysis category. How to operate the automatic report generator. The method of claim 10,
The analysis step,
When there is shared data, which is classification data shared between two or more analysis categories, a master analysis category having ownership for each shared data is designated so that refined data processed by processing the shared data can be generated only for the master analysis category. The method of operating the automatic report generation device, characterized in that. The method of claim 13,
The shared data,
A method of operating an automatic report generation apparatus, characterized in that ownership for data processing is allocated to any one of the two or more analysis categories based on at least one of a degree of relationship between data and a data size. The method of claim 13,
The shared data,
Data processing into refined data is performed during the data processing time specified between the two or more analysis categories,
The refinement data,
The method of operating an automatic report generation apparatus, characterized in that, when the data processing time ends or the data processing is completed, an analysis category that does not have ownership is updated. The method of claim 11,
The extraction step,
According to the content selection weight according to the core content extraction rule for each analysis category, the content of the change in the numerical physical quantity identified from the analysis data, the content of the numerical physical quantity compared between categories, and the specific term with the highest exposure frequency An operation method of an automatic report generation device, characterized in that extracting at least one of the contents related to the core contents. The method of claim 11,
The extraction step,
According to the time selection weight according to the core content extraction rules for each analysis category, at least one of the contents of recent changes rather than the past identified from the analysis data, contents of future prospects, and contents of the time before and after a specific event is the core. An operation method of an automatic report generation device, characterized in that extracting as content. The method of claim 11,
The extraction step,
When a hierarchical structure including a high-level category encompassing multiple categories of the same level is identified from the analysis data, the analysis content for the high level, multiple of the same level, according to the category selection weight according to the core content extraction rule for each analysis category. The content of the new upper-level category newly derived from the category, the content of the lower-level category showing a difference greater than or equal to the set value and the statistical physical quantity of the upper-level category, the category with the highest relevance to the analysis object among the categories of the same level The content of the category with the highest or lowest physical quantity among the categories of the same level, the content of the category with the physical quantity within the set rank or ratio among the categories of the same level, and the change of the physical quantity among the categories of the same level A method of operating an automatic report generator, characterized in that extracting at least one of the contents of the largest category as core contents.

Images