KR101985961B1 - Similarity Quantification System of National Research and Development Program and Searching Cooperative Program using same - Google Patents

Abstract

The present invention relates to a similarity quantification system for national research/development programs and a national research/development program cooperation suggestion method using the same. More specifically, according to the present invention, the similarity quantification system for national research/development programs comprises: a data construction module collecting a plurality of pieces of national research/development program information including a string about attribute information; a similarity calculation module calculating similarity between the collected national research/development programs; and a clustering module clustering the national research/development programs in accordance with their similarity. Accordingly, the similarity is calculated and quantized based on the string including the attribute information of numerous progressed or progressing national research/development programs, thereby providing an advantage of promoting cooperation between the national research/development programs.

Description

[0001] The present invention relates to a similarity quantification system and a national research & development collaborative proposal method using the same,

The present invention relates to a national research and development task similarity quantification system and a national research and development task collaboration method using the system. More particularly, the present invention relates to a national research and development task, The present invention relates to a system for quantifying a similarity of a national R & D task and a method for collaborating with a national R & D task using the same, which quantifies the similarity by comparing the strings included in the information, visualizes the similarity as a task group expected to be possible for future collaborations through cluster analysis.

A number of national R & D projects and tasks are being actively carried out in various fields. The information on these is being constructed in a single database, sharing and sharing the relevant information to enhance the efficiency of R & D investment and improve research productivity. It is being done steadily.

Especially, in recent years, due to the expansion of investment in disaster safety national R & D tasks, efficient R & D performance has been demanded both at home and abroad. In the United States and Japan, early on, they are aware of the complexity of disaster safety management and are establishing and promoting plans that emphasize collaboration among ministries. Respectively. Therefore, there is a need to provide an evaluation tool that can detect the possibility of collaboration between departments due to the characteristics of the disaster safety field.

However, most of the technologies that have been carried out so far focus on improving similarity, redundancy detection, and accuracy between tasks, and it is difficult to provide a fundamental answer to finding collaborative tasks.

To solve this problem, in Korean Patent Registration No. 1,426,765, a keyword is extracted from a table value of data to be analyzed for high-quality national R & D information provided by NTIS, and the expertise information of the person is derived based on the relational expression set for the extracted keyword Based on the attribute information and the link information for the expertise information of the derived person, the collaboration relation between the person and the person is derived as the knowledge map, and the expertise information of the person on the specific topic inputted by the keyword is determined, A system and method for providing a collaborative partner search service provided in the form of a knowledge map are known. However, this method is also applied to a disaster safety national research and development task having a relatively large complexity, Because of the limitations described above There is a problem that it is difficult to utilize it for finding collaborative tasks of national research and development tasks such as disaster safety field having relatively complexity.

Korean Patent Registration No. 1,426,765

In order to solve the above-mentioned problems, the present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to quantify the similarity of a task through not only a keyword but also a character string including attribute information of a national research and development project, cluster a plurality of tasks according to the quantified degree of similarity, The present invention aims to provide a system for quantifying similarity of national R & D tasks proposing efficient collaboration and a method of proposing collaboration between national R & D tasks using the same.

In particular, it is intended to meet the policy trend of disaster safety research and development in Korea and to provide a method to quantify the degree of similarity between R & D tasks and to support collaboration and review of future generations.

Other objects of the present invention will become more apparent through the preferred embodiments described below.

According to an aspect of the present invention, there is provided an information processing system including a data building module for collecting a plurality of national research and development task information including a string of attribute information, a similarity calculating module for calculating a degree of similarity between the collected national research and development tasks, It is a national research and development task similarity quantification system which includes a clustering performance module that clusters national R & D tasks. In this case, the national research and development task is a disaster safety research and development task, and the data building module classifies the attribute information of the national research and development task according to the category to constitute an information table of the corresponding national research and development task, And a cluster attribute analysis module for analyzing the national research and development task cluster clustered by the execution module by each cluster.

According to another aspect of the present invention, the degree of similarity is determined by inserting, deleting, or deleting a character string of a national R & D project information to be a reference and a string of a national R & And substitution. More specifically, the strings of the national research and development task information are p and p ', respectively, and are present between the p and p' The number of words is m and n, respectively, and δ is the number of times the editing operation was performed. That is, the edit-distance between p and p` strings, and when <sub>ED</sub> is similarity, the similarity of p and p` is D _ED (p, p`) = 1 -δ / (m + n ). &Lt; / RTI &gt;

In addition, the similarity calculation module is configured to construct an n × n symmetric matrix according to the number (n) of national research and development tasks for which the degree of similarity calculation has been completed. And the like.

According to another aspect of the present invention, in the proposed method of collaborating with national R & D tasks, the proposed method of national R & D task collaboration is performed by a national R & D task similarity quantification system, A data building module for collecting a plurality of national research and development task information including a string of information, a similarity calculation module for calculating the similarity between the collected national research and development tasks, and a clustering for grouping national research and development tasks according to the similarity The method comprising the steps of: collecting a plurality of national research and development project information including a string of attribute information by the data building module to construct national research and development project data; And the similarity calculation module Against a string of nine development project data similarity calculation step of calculating a similarity between the National Research Development Project; And a clustering step of clustering and visualizing national R & D tasks according to the degree of similarity by the clustering performance module.

In this case, the national research and development project may be limited to the disaster safety research and development project. The national research and development project similarity quantification system further includes a cluster attribute analysis module. The national research and development project collaboration method includes a cluster attribute analysis module And analyzing the cluster attributes according to whether or not the national research and development tasks are coincident with each other in the national research and development task cluster.

According to another aspect of the present invention, the similarity calculation module may include an insertion and deletion process to construct a string of a national R & D task information to be a reference and a string of a national R & deletion, and substitution, the degree of similarity between the national R & D tasks is calculated based on the minimum number of times the editing operation is performed. In more detail, Are p and p`, the number of words existing between the p and p` is m and n, respectively, and δ is the number of times the editing operation is performed. That is, the edit-distance between p and p` strings, and when <sub>ED</sub> is similarity, the similarity of p and p` is D _ED (p, p`) = 1 -δ / (m + n ). &Lt; / RTI &gt;

The similarity degree quantification system according to the present invention and the proposed method for collaborating with the national research and development project using the present invention can calculate the degree of similarity through a character string including attribute information of many national R & There is an advantage that collaboration between development tasks can be planned.

In particular, in the case of disaster safety national R & D tasks involving various ministries in the national R & D task, overcoming the problems of calculating similarity through simple keyword search, which generates issues on accuracy, It is possible to reduce the number of tasks to be analyzed in detail, and it is possible to carry out the suggestion of cooperation more smoothly through cluster analysis, and it is advantageous to meet the policy trend of the disaster safety research and development in Korea.

In addition, there is an advantage that the task can be grouped according to the degree of similarity and visualized so that it can be utilized in various fields such as research and development status and performance, in addition to the use of collaboration proposal.

1 is a block diagram of a system for quantifying the similarity of a national R & D task according to an embodiment of the present invention.
FIG. 2 is an example of an edit operation matrix for calculating the degree of similarity of a national research and development task according to an embodiment of the present invention; FIG.
FIG. 3 is a flow chart of a method for suggesting a national R & D task collaboration according to an embodiment of the present invention.
FIG. 4 is a visualized cluster diagram in a national R & D project collaboration proposal method according to an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention is capable of various modifications and various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the description. It is to be understood, however, that the invention is not to be limited to the specific embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof. Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram of a national research and development task similarity quantification system according to an embodiment of the present invention. Referring to FIG. 1, the detailed structure of a national research and development task similarity quantification system according to the present invention will be described below.

The national research and development task similarity quantification system according to the present invention includes a data building module (10) for collecting a plurality of national research and development task information including a string of attribute information and a national research development task A similarity calculation module 20; And a clustering module 30 for clustering national R & D tasks according to the degree of similarity.

And a cluster attribute analysis module 40 for analyzing the national research and development task cluster performed by the clusterization execution module 30 for each cluster.

In addition, the national research and development project referred to in the present invention can be applied to national research and development projects. In particular, it is desirable to apply the present invention to a task or a business having a high level of complexity because there are many related fields such as a disaster safety national research and development project.

The data building module 10 is connected to the National Science and Technology Information Information Service (NTIS), which has 628,350 national R & D project information data as of October 2017, so that it can utilize the national R & D project information .

At this time, the attribute information of the national R & D task constructed in the data building module 10 may be provided by the National Science and Technology Knowledge Information Service (NTIS), such as task name, performing agency, expected effect, task content, It consists of a string containing at least 60 pieces of attribute information or at least one of them.

Also, the data building module 10 classifies the attribute information of the national R & D task into categories such as task name, performing agency, expected effect, task content, summary of research content, keyword, .

FIG. 2 is an example of an edit operation matrix for calculating the degree of similarity of a national R & D task according to an embodiment of the present invention. Referring to FIG. 2, the degree of similarity between national R & The minimum number of times that the editing operation consisting of insertion, deletion, and substitution is performed in order to construct the string of the national R & D task information and the string of the national R & Is calculated on the basis of the reference value.

In other words, if the character string including 'ABCDEF' is included in the standard R & D task information and 'HBCEFG' is included in the national R & D task information to be compared, The following editing operation is performed.

An editing operation for substituting 'H' for 'A' is performed once, an editing operation for 'D' is inserted between 'C' and 'E' Deletion is performed once and a total of three editing operations are completed, a string including the national R & D task information to be compared and a string including one national R & D task information as the reference are the same . That is, the minimum number of editing operations is calculated through the editing distance algorithm as described above, and this corresponds to the degree of similarity between the national R & D tasks.

At this time, it is preferable that the string of each national R & D task for calculating the similarity is compared with the strings included in the same classification category among the task attribute information classified by the data building module 10.

The national R & D task in which the degree of similarity is calculated as described above constitutes an n × n symmetric matrix by the similarity calculation module 20 according to the number of national R & D tasks (n) as shown in Table 1 below. Through this, it is possible to cluster the national R & D tasks more easily in the clustering performance module 30 as well as to compare the similarity quantification between the national R & D tasks.

In the following Table 1, 'P' is a numeral or a letter for distinguishing tasks from each other such as the name or identification number of a national R & D project, 'N / A' means omission of calculation of similarity, '- Is the similarity value calculated by the module (20).

P (1) P (2) P (3) P (4) ㆍ ㆍ ㆍ ㆍ P (n) P (1) N / A
N / A

N / A


N / A



N / A




N / A P (2) - P (3) - P (4) - ㆍ
ㆍ
ㆍ ㆍ
ㆍ
ㆍ ㆍ
ㆍ
ㆍ ㆍ
ㆍ
ㆍ ㆍ
ㆍ
ㆍ P (n) - ㆍ ㆍ ㆍ ㆍ

As an example of the calculation of the degree of similarity by the similarity calculation module 20, the strings of the national R & D task information are p and p ', respectively, and the numbers of words existing between the p and p' are m and n And δ is the number of times the editing operation has been performed. That is, the edit-distance between p and p` strings, and when <sub>ED</sub> is similarity, the similarity of p and p` is D _ED (p, p`) = 1 -δ / (m + n ) &Lt; / RTI &gt;

For example, a string containing a standard national R & D project information and a string containing information on a national R & D project that is based on a string including the national R & In the case of p, if δ is 0, the similarity D _ED (p, p) between the national R & D tasks becomes 1.

On the other hand, the string p including the national R & D task information and the string p including the national R & D task information to be compared are compared with each other. As the number of words m and n of each word is constant and the editing distance δ becomes larger, The similarity D _ED (p, p) between national R & D tasks is close to zero.

That is, the similarity increases as the similarity degree D _ED (p, p) approaches 1, and the similarity decreases as the degree of similarity D _ED (p, p) approaches 0.

The clustering performing module 30 groups clusters of national R & D tasks according to the degree of similarity, and visualizes them by nodes and links.

In this case, the clustering performing module 30 is an open source program that enables data searching and processing. It can efficiently visualize a complex network and analyze various clustering or network analysis tools Can be utilized.

In addition, the cluster attribute analysis module 40 may confirm the correspondence of the main keywords of the national research and development tasks according to the clusters, and may further constitute a national research and development task group having key keywords in the clusters.

FIG. 3 is a flowchart of a method of suggesting a national R & D task collaboration according to an embodiment of the present invention. Referring to FIG. 3, the detailed structure of a national R & D task collaboration method will be described below.

In the national R & D task collaboration proposal method, the national R & D task collaboration suggestion method is performed by a national research and development task similarity quantification system, and the national research and development task similarity quantification system includes a plurality of A similarity calculation module 20 for calculating the similarity between the collected national research and development tasks, and a clustering execution module 30 for clustering national research and development tasks according to the similarity The national research and development project collaborative proposal method includes collecting a plurality of national research and development project information including a character string of attribution information by the data building module 10 to construct national research and development project data Building step S10; A similarity calculation step (S20) of calculating the similarity between the national R & D tasks by comparing the string of the national R & D task data by the similarity degree calculation module (20); And a clustering step (S30) of clustering and visualizing national R & D tasks according to the degree of similarity by the clustering performing module 30 (S30).

In particular, the national research and development project is characterized by disaster safety research and development that is directly related to national security, structuring relatively complex problems that arise in relation to various departments, not related to one department.

The national research and development task similarity quantification system further includes a cluster attribute analysis module 40. The national research and development task collaboration method is implemented by a cluster attribute analysis module 40 in a national R & And a cluster attribute analysis step S40 for classifying the keyword according to the keyword matching.

In addition, the similarity calculation module 20 inserts, deletes, and deletes character strings of a national R & D project information and a string of national R & Based on the minimum number of times the editing operation, which consists of substitution, has been performed.

More specifically, the strings of the national research and development task information are respectively p and p`, the number of words existing between the p and p` is m and n, respectively, and δ is the number of times the editing operation is performed. That is, the edit-distance between p and p` strings, and when <sub>ED</sub> is similarity, the similarity of p and p` is D _ED (p, p`) = 1 -δ / (m + n ), And it is performed in the same manner as the above-described similarity degree quantification system of national research and development task, and thus a detailed description thereof will be omitted.

The cluster attribute analysis step (S40) is preferably performed only on national R & D project communities having the calculated average similarity value of 0.2 to 0.6, rather than being performed on all the clusters.

In addition, the cluster attribute analysis step S40 may be performed only for a cluster having a similarity average value that is higher than the average similarity value between the calculated national R & D tasks according to the character string to be subjected to the similarity extraction and the characteristics of the task.

The practical application example of the national research and development project collaboration method constituted as described above will be described as follows.

(S10) for extracting National R & D tasks 91 in the disaster safety field conducted from 2013 to 2017 by a data building module 10 connected with the task search system of National Science and Technology Knowledge Information Service (NTIS) Respectively.

Thereafter, the similarity calculation module 20 applies the character string corresponding to the summary of the research content among the strings including the attribute information of the national research and development project of each disaster safety field extracted in the above, A calculation step S20 was performed. At this time, it is desirable that the similarity of the national R & D tasks is arranged in a 91 × 91 symmetric matrix according to the number of national R & D tasks in the disaster safety field extracted as shown in Table 1 above.

As a result of examining the similarity between the national R & D tasks extracted from the above, the similarity was calculated to be 1 for the same task, and the 7 tasks approximated to 1 were analyzed. As a result, Was derived.

Then, the clustering is performed according to the similarity value through the clustering step S30 performed by the clustering performing module 30, and the clustering is performed using various tools such as a histogram, a matrix, and a graph.

FIG. 4 is a visualized cluster chart in the proposed method of collaborative research project according to an embodiment of the present invention. In this example, 91 national R & D tasks were classified into a total of 21 clusters. More specifically, P01 to P91 shown in FIG. 4 are serial numbers assigned to distinguish each task, which means that the task belongs to an elliptic cluster. Through this, it is possible to know which national R & D project is included in the community.

The average value of the sample was 0.18, the median value and the mode value were 0.19, and the median value and the mode value were 0.19. As shown in Table 2, Can be classified and visualized.

Cluster No. Ave Cluster No. Ave Cluster No. Ave 11 1.00 19 0.29 12 0.24 4 0.99 15 0.28 9 0.23 6 0.98 3 0.27 18 0.23 17 0.53 2 0.27 10 0.23 8 0.53 16 0.27 13 0.22 7 0.30 14 0.26 5 0.19 20 0.30 One 0.25 0 0.18

In Table 2, 'Cluster No.' is a serial number for distinguishing clusters, and 'Ave' is an average value of similarities of the clusters. When the threshold value according to the similarity quantification from the result is about 0.2 or more, it is analyzed that it is possible to obtain explanatory power about the similarity between the tasks, and it is possible to derive the tasks that can be collaborated with the tasks having the similarity degree higher than the average.

Then, the cluster attribute analysis module 40 analyzes the keywords of the tasks included in the first cluster by performing the cluster attribute analysis step (S40). As a result of analyzing the keywords of the tasks included in the first cluster, 'Yes alarm', 'Disaster situation management' , 'Intelligent CCTV', and so on, and it was possible to seek collaboration with 'development of monitoring system through building data'. The possibility of similar collaboration was confirmed in the 7th cluster. The cluster was identified as 'satellite', 'drought', 'remote exploration', 'disaster information', and 'spatial information'. I could. Also, it is analyzed that it includes the possibility of cooperation between 1 cluster and 7 cluster.

It will be apparent to those skilled in the relevant art that various modifications, additions and substitutions are possible, without departing from the spirit and scope of the invention as defined by the appended claims. The appended claims are to be considered as falling within the scope of the following claims.

10. Data building module
20. Similarity calculation module
30. Clustering performance module
40. Cluster Property Analysis Module
S10. Steps to build data
S20. The similarity calculation step
S30. Clustering step
S40. Cluster attribute analysis step

Claims (14)

A data building module for collecting a plurality of national research and development project information including a character string for attribute information and classifying the attribute information of the national research and development project by category and constructing an information table of the corresponding national research and development project;
A similarity calculation module for calculating a similarity between the collected national R & D tasks and constructing a symmetric matrix of n × n according to the number (n) of national research and development tasks for which similarity calculation is completed;
A clustering module for clustering national R & D tasks according to the similarity; And
And a cluster attribute analysis module for checking and classifying the keyword matching of the national research and development task by the cluster of the national research and development task grouped by the clustering performance module,
The national research and development task is disaster safety research and development task ,
The degree of similarity consists of insertion, deletion, and substitution in order to construct a string of one National R & D task information and a string of national R & D task information to be compared. Wherein the calculation is performed based on the minimum number of times the editing operation has been performed,
The string of the national research and development task information is p and p`, the number of words existing between the p and p` is m and n, respectively, and δ is the number of times the editing operation is performed. That is, the edit-distance between the p and p` strings, and when the DED is similarity,
The similarity degree between p and p 'is calculated using the equation of DED (p, p`) = 1-δ.
delete delete delete delete delete delete delete In the National R & D Task Collaboration Proposal Method,
The national research and development task collaboration method is performed by the national research and development task similarity quantification system,
The national research and development task is disaster safety research and development task,
The national research and development task similarity quantification system includes a data building module for collecting a plurality of national research and development task information including a string of attribute information, a similarity calculation module for calculating the similarity between the collected national research and development tasks, And a cluster attribute analysis module for classifying national research and development tasks according to whether or not the national research and development tasks are coincident with each other by each cluster of national research and development tasks clustered by the clusterization execution module,
The National Research & Development Project Collaboration Proposal Method
A data constructing step of constructing national R & D task data by collecting a plurality of national R & D task information including a character string of attribute information by the data building module;
In order to calculate the similarity between national R & D tasks by comparing the string of national R & D task data by the similarity calculation module, the string of the national R & D task information to be a reference and the national R & The similarity between the national R & D tasks is calculated based on the minimum number of times the editing operation, which includes insertion, deletion, and substitution,
The strings of the national research and development task information are respectively p and p`, the number of words existing between the p and p` is m and n, respectively, and δ is the number of times the editing operation is performed. That is, the edit-distance between the p and p` strings, and when the DED is similarity,
The degree of similarity between p and p 'is calculated using the equation of DED (p, p`) = 1-delta;
Clustering the national R & D tasks according to the degree of similarity by the clustering module and visualizing them; And
Analyzing attributes of each cluster of the clustered national R & D tasks by the cluster attribute analysis module, analyzing cluster attributes only for the cluster having a similarity average value that is higher than the average value of the calculated similarities among all the national R & The National R & D Project Collaboration Proposal Method. delete delete delete delete delete

Images