WO2018212398A1 - Mind-mining analysis method using link between view data - Google Patents

Abstract

The present invention relates to a method for performing a mind-mining analysis using a link between view data in an analysis server connected via a network with a plurality of user terminals and participant terminals. The method comprises the steps of: receiving information related to issue information and target information from the user terminals and generating a virtual room field associated therewith; when receiving view data regarding the issue and the target from the user terminals or the participant terminals, loading the view data in the virtual room; when receiving, from the user terminals or the participant terminals, information on a link connecting the view data and connection view data regarding a reason for establishing the link, loading the information and the connection view data in the virtual room; and establishing a path connecting the issue information and the target information on the basis of the order of a logical flow of the view data and frequency information of links connecting the view data. According to the present invention as described above, by using mind-mining, it is possible to identify a risk by structuring a problem from an uncertainty triggered by a current change factor and thereby to derive a solution or alternative through management techniques such as monitoring, visualization, simulation, etc.

Description

Mining Mining Analysis Method Using Linkage Between Perspective Data

The present invention relates to a mind mining analysis method using a link between a view data presented on an issue of interest and a presented goal or expectation, and more specifically, presented by various participants with respect to a specific issue. Link structure analysis between perspective data and perspective data can be used for problem structuring for decision making, risk analysis such as weak signal detection, hypothesis setting for targets through scenarios, and design and verification activities. It relates to an analysis method.

Big data is data generated in a digital environment. It is large in size, short in generation cycle, and includes large-scale data including text, pictures, and video data as well as numerical data. Digital environment such as location information and social network service (SNS) has increased the amount of data and the variety of data types compared to the past, providing an environment for analyzing and predicting not only people's behavior but also thoughts and opinions.

Big data is being produced not only in the private sector but also in the public sector. A large amount of data is produced in the fields of Census, various social surveys, national tax data, medical insurance and pensions. This diverse and vast amount of data is attracting attention because it can be used as an important resource that determines the competitiveness of the future.

Recently, there has been an increasing demand and research for unstructured big data analysis methods. Unstructured data, unlike numeric data, is unstructured data that is complex in form and structure, such as text, pictures, images, and documents. Atypical data such as blogs, messengers, bulletin boards, etc., which are exploding on the web, are attracting attention because they can grasp the flow of public opinion. Unstructured data analysis methods include text mining, web mining, and opinion mining.

Big data analysis is accomplished through data mining, which combines machine learning, artificial intelligence, and computer science with traditional statistical methodologies. . In other words, not only statistical methodologies but also machine learning & deep learning, neural network analysis, etc., are being used in various approaches to extract information from data.

Among big data analysis, web mining and opinion mining techniques are mainly used for analyzing unstructured big data.

Web mining refers to data mining for the web that extracts useful information from web log information or search words generated during the process of using the Internet.

Opinion mining refers to analyzing people's opinions, evaluations, attitudes, and feelings about an issue, person, issue, or event. Analyze sentences by gathering people's subjective opinions on specific issues. In sentence analysis, the facts and opinions are divided, the opinions are extracted, divided into positive and negative, and the strength is measured.

Opinion mining uses automated analysis because it is large-scale web documents such as portal boards, blogs, and shopping malls. Opinion mining also uses natural language processing (NLP) and computational linguistics, which are used in text mining, because it is the text to be analyzed.

However, the conventional big data analysis method using artificial intelligence, etc. have the following problems.

First, data input for analysis may be a priori error to which the less relevant data is refined. For example, a large number of data, which are considered to be less relevant from the current point of view and are excluded from the analysis (weak signal: weak signal), become more important over time, so that wild card and emerging risks are emerging. A priori error can occur in the process of data cleansing, as informational cases can occur.

Second, opinion mining among unstructured data mining techniques distinguishes positive and negative opinions by dividing opinions into positive and negative to measure the intensity of the opinions. There is not enough time to use it as a tool for decision making.

Third, before or after the application of an automated analysis method such as artificial intelligence, the analyzed data is clustered among related data to pattern or normalize the analysis result data. This is done through the subjective interpretation process of the subjects, and in some cases, the experts look at the data differently and can interpret differently.

Therefore, in big data analysis, there is an increasing demand for a methodology that can solve the problem of a priori error in the process of selecting data to be analyzed and the analysis data patterning by subjective interpretation of a few experts.

The present invention has been proposed to solve or supplement the above problems, and an object of the present invention is to provide a strength and a connected bridge of link (bridge) connection information connecting view data of a plurality of participants. ) The human learning technique that uses mind mining analysis using data enables big data analysis from a more objective point of view.

Another object of the present invention is to collect weak signals having low connection strength and monitor and analyze their trends or flows to improve data reliability in the analysis process.

According to a preferred aspect of the present invention for achieving the above object of the present invention, a method for performing a mind mining analysis using a link between perspective data in an analysis server connected to a plurality of user terminals and participant terminals via a network The method may further include: receiving information related to issue information and target information from the user terminal, generating a virtual room field related thereto, and when receiving viewpoint data about the issue from the user terminal or a participant terminal, loading the information into the virtual room. Doing;

Receiving information on a link connecting the viewpoint data from the user terminal or a participant terminal and connection viewpoint data on a reason for establishing the link, loading the same into the virtual room, and logical flow order of the viewpoint data And setting a path connecting the target information from the issue information based on frequency information of links connecting the viewpoint data with each other. do.

Here, the step of setting the path connecting the target information from the issue information based on the logical flow order of the viewpoint data and the frequency information of the links connecting the viewpoint data, the link having a high frequency as a priority It may be to set a path connecting the target information from the issue information, it is preferable that the priority is given to a link of the position close to the issue information in view of the logical flow of the viewpoint data.

The viewpoint data and the connection viewpoint data may include at least one of unstructured data and text data, and the text data may be a word or sentence for describing the atypical data.

In addition, the analysis server may generate a path description for each link by performing text mining on the perspective data constituting each link and the text data included in the connection perspective data, and further, each link constituting the established route. Text mining may be performed on the path descriptions for to generate a comprehensive description of the entire path.

Here, the text mining may be performed including a semantic data refinement step, a main part-of-speech word extraction and frequency calculation step, and a part-of-speech word rearrangement step.

The analysis server extracts, as a weak signal, viewpoint data belonging to a link having a link frequency lower than a reference value, and detects a periodic increase level of the weak signal, and when the increase level exceeds a threshold, the viewpoint belongs to the weak signal. It is also possible to set the path including the data as the path to be analyzed.

According to the present invention as described above, it is possible to identify the prognosis and risks such as weak signals by structuring the problem from the uncertainty triggered by the current change factors using mind mining, and using the management techniques such as monitoring, visualization and simulation This has the effect of deriving solutions or alternatives.

The present invention has the advantage of high usability in various fields such as problem structuring for decision making, risk analysis such as weak signal, hypothesis setting for target through scenario derivation, and design and verification activities.

1 illustrates a configuration of a mind mining analysis system according to the present invention.

FIG. 2 illustrates a concept of a mind mining analysis method using a bridge bridge between view data presented about an issue and a goal or an expectation.

3 is a flowchart illustrating a path description generation procedure.

4 is a flowchart illustrating a weak signal detection and analysis method.

5 illustrates a visualization of the results of mind mining analysis.

Hereinafter, with reference to the accompanying drawings will be described in detail the present invention. It should be noted that the same elements in the figures are denoted by the same reference signs wherever possible. On the other hand, the terms or words used in the present specification and claims are not to be construed as limiting the ordinary or dictionary meanings, the inventors should use the concept of the term in order to explain the invention in the best way. It should be interpreted as meanings and concepts corresponding to the technical idea of the present invention based on the principle that it can be properly defined. Therefore, the embodiments described in the present specification and the configuration shown in the drawings are only the most preferred embodiments of the present invention, and do not represent all of the technical ideas of the present invention, and various alternatives may be substituted at the time of the present application. It should be understood that there may be equivalents and variations.

1 illustrates a configuration of a mind mining analysis system according to the present invention. As shown in FIG. 1, the mind mining analysis system according to the present invention may include a big data platform 10, an analysis server 20, a user terminal 30, and a participant terminal 40.

The big data platform 10 is an IT environment required to process and analyze data collected from various data sources, extract knowledge, and provide intelligent services based on the big data platform. You need to have the ability to collect and integrate heterogeneous data, access and process data quickly, store and manage large amounts of data, and analyze large amounts of heterogeneous data to the desired level.

The big data platform 10 may be, for example, a text mining service such as a news big data platform of the Korea Press Foundation, and may visualize big data or provide information with high frequency through text mining. The mind mining analysis method according to the present invention may be performed by selecting high frequency information as an issue and presenting a goal or expectation about the issue.

In this embodiment, but setting the issue for the mind mining analysis through the big data platform, but without the information from the big data platform, users arbitrarily set their own issues, goals or expectations, a large number of participants It is also possible for a mind mining analysis according to the present invention to be made by presenting a view of the issues, goals and expectations.

Analysis server 20 is a view that a plurality of users and participants through the user terminal 30 and the participant terminal 40 can be present on the issue presentation, goal and expectations presentation, the relationship between the issue and the target (view ) Provides virtual space where activities such as presentation, bridge (connection) connection creation, and connection bridge (connection bridge) presentation for link formation can be performed, and creates a virtual room corresponding to the number of issues. It is possible to do this or to present several issues that are related to one virtual room. The analysis server 20 generates a virtual room field in the database in response to a request signal for opening a virtual room of the user terminal 30.

The analysis server 20 receives from the user terminal 30 or the participant terminal 40 a plurality of viewpoint data on an issue and a goal, and path data and connection view data for a link formed through a link connection between the viewpoints. This data is then loaded onto the virtual room so that all participants can see the information uploaded and perform mind mining analysis based on the received data.

The user terminal 30 is a communication terminal of a user who sets an initial event for a specific issue and presents his or her desired goal, and is a PC, a laptop, a tablet PC, a smart phone, and other display devices (eg, virtual reality and augmented reality). Various communication terminals such as) may be used. The user may set an initial event as a change factor or a trigger event for an issue provided through the big data platform 10 or set an initial event for an issue regarding a change factor that he or she thinks. The user may further set up any intermediate events that are expected before the goal or expectation from the change factor.

The user can present his or her desired goals or expectations, and as a participant he / she can join a virtual room opened with issues he has selected or presented.

Participant terminal 40 is a communication terminal used by a participant participating in a virtual room opened by a user, and various communication terminals such as PCs, laptops, tablet PCs, smart phones, and other display devices (eg, virtual reality and augmented reality). This can be used. Participants participate in virtual rooms created by users, presenting their perspectives by uploading unstructured data based on issues and anticipated events and goals or expectations, or creating links that can involve unstructured data between views. Present your own explanation (ie, connection perspective).

2 is for explaining the concept of the mind mining analysis method using a link between the perspective data presented for the issue.

The user connects to the analysis server 20 through the user terminal 30 and establishes a virtual room in which a large number of participants can participate in their own issues and goals or expectations, and is interested in those issues and goals or expectations. A plurality of participants may access the analysis server 20 through the participant terminal 40 and participate in the established virtual room to present a viewpoint on the corresponding issue, goal or expectation.

For example, as an example of a mind mining analysis of political issues, the issue may be related to "Who is eligible to be president in this presidential election?" Set as information, and as goals and expectations, "○○○ should be president." And the like can be set. Just as there may be a plurality of issues in one virtual room, there may be a plurality of goals or expectations in a virtual room.

In FIG. 2, the viewpoint data 100 is information about a viewpoint presented by a user or a participant with respect to a corresponding issue, a goal, or an expectation, and may include descriptive data 110 and descriptive text 120 about the unstructured data 110. Can be. The unstructured data 110 may be a picture, a video, a URL, a document, etc. related to the issue, and the description text 120 may be a sentence explaining a reason for uploading the unstructured data in relation to the current issue.

The connection viewpoint data 200 is used to present a description of a participant linked and linked between the viewpoint data 100, and may include descriptive text 220 for the unstructured data 210. The unstructured data 210 may upload a picture, video, URL, document, etc. related to the issue, and the description text 220 may be a sentence explaining a reason for generating a corresponding link in relation to the current issue and goal. .

The connection viewpoint data 200 further includes a connection line 300 having directivity as connecting the viewpoint data 100. For example, in FIG. 2, the connection line 300 between the viewpoint data A 100 and the viewpoint data B 100 is a participant when the viewpoint A and the viewpoint B are related to each other or the two viewpoints are combined with respect to the issue and the target. If it is determined that it is good to link between them, any number of the selected links can be used as long as the number of selected links to make the connecting line thicker or the color of the connecting line to recognize the link frequency.

Referring to FIG. 2, first, a user may set up an initial event by presenting a goal and an expectation of a virtual room on the subject of his / her interest, and presenting his viewpoint data A on the issue of interest. Participants can then present their views (perspectives B, C, D, etc.) on the issues and goals. Here, the viewpoint may be an opinion between the relationship of the opened issues and the goals or an opinion on a specific issue and goal. As described above, the viewpoints presented herein may be unstructured data 110 and descriptive text 120 thereof, or may include only one of them. In this case, for the text mining analysis, the explanatory text is preferably presented.

Participants may also present linking perspective data (Link AB, Link BD, Link AC, Link CD) without presenting the viewpoint or linking between the viewpoint data and the viewpoint data with the viewpoint presentation. Here, the connection point of view may be a comment or explanation on why the participant connects the two points of view, and may be related to a case in which the two points of view may be a good opinion or a good answer. As described above, the connection viewpoint data 200 presented here may be the unstructured data 210 and the description text 220 thereof, or may include only one of them. In this case, it is preferable that the explanatory text is presented for the text mining analysis.

In FIG. 2, the analysis server connects each link to set a path starting from an issue of interest and reaching a goal. In this case, a plurality of paths connecting the links may be formed, and in FIG. 2, a path connecting the viewpoint A-a viewpoint B-the viewpoint D and a path connecting the viewpoint A-the viewpoint C-the viewpoint D, that is, two paths exist. The case is illustrated.

In this case, a rule forming a path connecting the issue and the target of interest may be set as follows.

(1) The frequency of the link, that is, the higher the number of participants who select the link, the higher the priority, and linking the high-priority links, one path is completed.

(2) Priority is given to links among the links that are close to the issues presented (links that are located close to the issues presented in the logic flow). This means that if there are a lot of links in the view presented earlier in the logic flow, the view is preempting the issue first, and the priority may be different if analysis from other points is required. Here, the logic flow may be an order in time flow.

For example, in FIG. 2, if the frequency of the connection line of the link AB and the link CD of the link CD is high, and the connection line of the link BD and the link AC is relatively low in frequency, applying the above rule, the high frequency link AB and the link CD are applied. Has a priority, but since the link AB is located at a faster point in the logical flow than the link CD, the path connecting the viewpoint A-the viewpoint B-the viewpoint D including the link AB is selected as the main route and analyzed This is done.

3 is a flowchart illustrating a path description generation procedure.

In FIG. 2, when a plurality of viewpoint data and connection perspective data are presented by a plurality of participants, it is necessary to visualize and provide them.

As a method of visualization, when the number of viewpoint data and the connection viewpoint data is small, all of them can be displayed.However, when the number of viewpoint data and the number of viewpoint data is large, the text mining for them can be used to represent a sentence or a symbol having representative keywords or representative keywords ( Symbol, etc.).

2 illustrates a method for this, when a plurality of viewpoint data and connection perspective data are collected, the analysis server 20 understands the contextual meaning of the machine viewpoint data and the connection viewpoint data through the semantic data analysis method, Highly relevant data are selected through inference and purification (S300).

When the data is purified through the above process, the analysis server 20 extracts the words for each of the main parts of speech that are commonly used in the viewpoint data, and calculates the frequency of the words (S310). As the main parts of speech, adjectives, nouns, and verbs can be used as important parts of speech in conveying meaning.

Next, the analysis server 20 rearranges the words of the main parts of speech having a high frequency to generate a path description (S320).

For example, in the step S310, if "substantial" as a high frequency adjective, "ERWS" as a noun, "acceptability", and "investigate" as a verb are extracted, the path description is performed as "investigate substantial ERWS". in the form of "acceptability". Through this process, two viewpoint data constituting one link and connection viewpoint data describing the connection reason thereof can be represented by one or a plurality of path descriptions.

Furthermore, the issues and goals presented are compressed through the analysis of text mining for the path descriptions of each link that is connected through a path having a link structure of links with multiple logical flow sequences. It may also be possible to generate descriptive statements.

4 is a flowchart illustrating a weak signal detection and analysis method.

As described in the background art, the data input for analysis may have some a priori error as the data having low relevance is refined, and FIG. 4 provides a method for solving the problem.

In FIG. 2, the viewpoint data 100 having no connection line 300 or having a link to which the connection line 300 goes is information that is not used in the mind mining analysis. Therefore, the analysis server 20 regards such data as a weak signal and extracts and stores only the weak signal (S400).

The analysis server 20 detects an increase level of the weak signals by analyzing the frequency of the weak signals at regular intervals (S410).

The analysis server 20 analyzes the trend of the weak signal based on the increase level information of the weak signals and generates a report (S420). When the increase level of the weak signal exceeds the threshold, the analysis server 20 may consider the weak signal as a wild card and perform a mining mind analysis reflecting the corresponding viewpoint data.

For example, as described with reference to FIG. 2, when the frequency of the connection line of the link AB and the link CD of the link CD is high, and the connection line of the link BD and the link AC is relatively low, the viewpoint A-the viewpoint C-the view D are connected. The path is recognized as a weak signal.

Therefore, the mining mind analysis is not performed on the path, but the information is continuously extracted and stored as the weak signal, and the signal is also meaningful if the increase level of the weak signal is detected at a predetermined period and the threshold is exceeded. Recognized as, the mind mining analysis is also performed on the path corresponding to the weak signal to enable more objective analysis.

5 illustrates a visualization of the results of mind mining analysis.

In FIG. 2, when the number of viewpoint data and the connection viewpoint data is large, it may be efficient to visualize the mind characteristics of the corresponding data in a single icon or symbol form. This visualization is a kind of system dynamics simulation technique. The embodiment of FIG. 5 illustrates a case in which the mind characteristics of each viewpoint are represented by a picture of the life cycle of the planet.

Although the present invention has been described in connection with the above preferred embodiments, it is possible to make various modifications or variations without departing from the spirit and scope of the invention. Accordingly, the appended claims will cover such modifications and variations as fall within the spirit of the invention.

Claims (9)

In the method of performing the mind mining analysis using a link between the viewpoint data in the analysis server connected to a plurality of user terminals and participant terminals through a network, Receiving information related to issue information and target information from the user terminal and generating a virtual room field related thereto; Loading viewpoint data of the issue from the user terminal or the participant terminal into the virtual room; Loading information on a link connecting the viewpoint data from the user terminal or the participant terminal and connection viewpoint data on a reason for establishing the link, loading the information into the virtual room; And setting a path for connecting the target information from the issue information based on the logical flow order of the viewpoint data and the frequency information of the links connecting the viewpoint data. Mining Mining Analysis Method Using. The method of claim 1, The setting of the path connecting the target information from the issue information based on the logical flow order of the viewpoint data and the frequency information of the links connecting the viewpoint data may include: Mining analysis method using a link between the perspective data, characterized in that for setting the path connecting the target information from the issue information. The method of claim 2, A priority mining analysis method using a link between viewpoint data, wherein priority is given to a link located near the issue information in view of logical flow of the viewpoint data. The method of claim 2, And the viewpoint data and the connection viewpoint data comprise at least one of unstructured data and text data. The method of claim 4, wherein And the text data is a word or sentence for describing the atypical data. The method of claim 4, wherein The analysis server generates a path description for each link by performing text mining on the viewpoint data constituting each link and the text data included in the connection viewpoint data. . The method of claim 6, And the analysis server performs text mining on the path descriptions of each link constituting the set path to generate a comprehensive description of the entire path. The method according to claim 6 or 7, The text mining method includes a semantic data refinement step, a main part-of-speech word extraction and a frequency calculation step, and a part-of-speech word rearrangement step. The method of claim 4, wherein The analysis server extracts viewpoint data belonging to a link whose link frequency is lower than a reference value as a weak signal, detects a periodic increase level of the weak signal, and extracts the viewpoint data belonging to the weak signal when the increase level exceeds a threshold. Mining analysis method using a link between the perspective data, characterized in that the path to be included as an analysis target path.

Images