JP2020035077A - Technical capability determination support system - Google Patents

Abstract

To provide a program which supports determining person's skill.SOLUTION: A program is provided which causes a computer to implement: a function of acquiring data relating to a repository held by a user from a database; a function of digitizing the data relating to the repository on the basis of at least a result of code parsing; a function of storing a parameter associated with the data relating to the repository, which is indicative of dependency of the user on other users; a function of learning a relation between a numeric value and a skill deviation of the user by machine learning; a function of updating weighting coefficients among respective layers of a learned neural network on the basis of the parameter; and a function of using the learned neural network to calculate a skill deviation of the user.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a system for supporting the skill determination of a person for a specific use, and more particularly to a system for supporting a skill diagnosis of a person for a specific use through appropriately evaluating a data source. .

  Programming is an example of an application-specific person's skill. Conventionally, there has not been a technique for scoring whether or not programming is possible, but there is, for example, a technique for analyzing and scoring the technical ability concerning programming using an evaluation concerning programming as shown in Patent Document 1. . The so-called technical ability can be measured by performing an evaluation in programming.

  However, in performing skill determination of a person such as programming, even if a skill diagnosis is performed using a technique as in Patent Document 1, it is not always possible to support the skill diagnosis of a specific person. This is because skill diagnosis has a judgment purpose, and there is a specific diagnosis method suitable for the purpose. Also, a skill determined using a technique as in Patent Literature 1 does not always give motivation to a person (a job seeker in the case of job change).

JP 2015-146176

  In determining the skills of a person, the person in charge must rely on the experience and intuition of the person in charge (in the case of a job posting, the person in charge of human resources) to judge the person's skills and provide advice to the person in charge. There was no mechanism to specify job seekers' skills.

  According to one aspect of the present invention, there is provided a system for automatically performing a skill determination of a person and providing visualization of one's technical ability by providing feedback of the technical ability to the person and providing scored data. More specifically, it is an object of the present invention to provide a system that supports skill determination of a person by evaluating input technical information from a plurality of viewpoints from the viewpoint of a specific application.

  According to the present invention, in a support system for skill determination of a specific person, input means for acquiring predetermined information data of a specific skill, and skill information are analyzed based on a skill information evaluation rule, and the input information is analyzed. At least one of the programming language used in and the influence on social is extracted, and the influence on the extracted programming language or social is analyzed based on the technical ability determination rule. A skill analysis means support system for generating an evaluation score of 2.

  Anyone can properly make skill judgments according to their technical skills without relying on the experience and intuition of a person's skill diagnosis person.

It is a figure showing the functional block of the technical ability judgment support system of this example. It is a figure showing an example of an analysis part of a technical ability judgment support system of this example. It is a figure showing an example of an analysis part of a technical ability judgment support system of this example. It is a figure which shows the display part of the technical ability determination assistance system of a present Example. It is a figure which shows the display part of the technical ability determination assistance system of a present Example.

  One of the visualizations of a specific person's technology is open source. In the skill judgment of a person, there is a difference in the skill for each person, and especially in open source data with a lot of skill information, it is not possible to measure the effective technical ability including the collection of information.

  In the present embodiment, an open source analysis scoring system that provides an open source analysis service will be described as an example of a technical ability judgment support system. This open source analysis and scoring system is a system used to judge a person's skill from the influence on social sources provided along with open sources on the Internet.

  When the user inputs the open source account information to the user terminal, a request jumps from the user terminal to the database, and data processing and data storage are performed on the database. . As a result of the analysis on the database, the result is displayed on the user terminal, and the user can obtain the analysis result in the programming language and advice on the analysis result.

  FIG. 1 is a conceptual diagram of a technical skill determination support system 1000. The technical ability judgment support system 1000 includes an input unit 1010, an input information analysis unit 1020, an input comparison analysis unit 1030, a score calculation unit 1040, an input information or analysis result display unit 1050, an input information evaluation rule database 1110, and an input comparison analysis. A rule database 1120, a score calculation database 1130, an input information and analysis result storage database 1140, and a control unit 1200 are provided.

  The user terminal 1300 is a device for displaying a user's technical skill determination result, which will be described later, by displaying an open source account on the Internet and technical information of the programming language. The user terminal 1300 may be provided with a function capable of displaying account information open to the public and a predetermined display, and may be a personal computer, a tablet terminal, a smartphone, or any type. In this embodiment, the user terminal 1300 is arranged outside the technical skill judgment support system 1000. However, as another embodiment, the user terminal may be arranged inside the technical skill judgment support system 1000. .

  The input unit 1010 obtains account data input from the user terminal 1300.

  The input information analyzing unit 1020 reads the open source data 1400 of the account based on the account data obtained from the input unit 1010. Here, examples of the types of input data are (1) to (5) below.

(1) Pull Request including Issue, Merge availability, and Proposal Content: The language handled in the pull request based on the pull request, the degree of influence such as whether the pull request has been captured, and the like. You can determine how good the repository has been affected.

(2) Organization data created in GitHub: Based on the data, it is possible to determine an organization level such as what kind of user the organization is composed of and what repository the organization has.

(3) README and source code data: The language of the repository and the quality of the code in the repository can be determined based on the data.

(4) User's own data, quantitative data such as the number of stars and commits in the repository: quantitative information itself such as the number of commits, the number of pull requests, and the number of followers provided by APIs

(5) Follow information between users: A quantitative value of a personal connection can be determined based on the information.

  The data described above is a state in which data such as "who", "what kind of information", and "when it was created" are arranged and stored. It should be noted that data judgment such as good or bad for the data at this time is not performed.

  After reading the open source data, the input information analysis means 1020 is decomposed into a plurality of pieces of information, and the decomposed information is analyzed by the analysis device means or quantitative information itself is used.

(1) A flow of a pull request including whether or not an issue or merge is possible and a proposal content will be described.
The input information analysis means 1020 provides a programming language and other open source using a programming language based on a proposal for improvement to the open source in the existing programming language in the open source data and provision of the improvement proposal contents or information. Analyze the degree of impact.

  The quantitative information and the quantized information analyzed by the pull request analysis device are stored in the input information and analysis result storage database 1240. Table 1 shows an example of a data structure stored in the analysis result storage DB 1240.

(2) The flow of data of an organization created in GitHub will be described.
Based on the community to which the relevant account belongs in the open source data, the organization analysis device analyzes the organization analyzed in advance, or the level of the community or organization, and the degree of impact of the organization.

  The quantitative information and the quantized information analyzed by the tissue analysis device are stored in the input information and the analysis result record database.

(3) The flow of data of README and source code will be described.
Based on open information such as source code and README released by the person concerned, a programming language technology analyzer analyzes the effects of a pre-analyzed programming language or its level in the programming language and the clarity of sentences. .

  The quantitative information and the quantized information analyzed by the programming language technology analyzer are stored in the input information and analysis result record database.

(4) Quantitative data flow such as user data itself and the number of stars and commits in the repository will be described.
The information provided by the open source information data source is stored in the input information and analysis result record database.

(5) The flow of follow information between users will be described.
Answers in open source data Based on the account or organization that the account follows or is followed by the personal connection analyzer, the account or organization or the level of the account and organization, the impact of the account or organization Analyze the relationship based on.

The quantitative information and the quantized information analyzed by the personal connection analyzer are stored in the input information and the analysis result record database.

  The input information and the data stored in the analysis result record database are arranged.

  Based on the data arrangement, the integrated score of each language in the account is performed in the analysis part of the technical ability judgment support system. Here, a numerical result is obtained using a calculation model created by machine learning.

  The advice of the judgment result is created in the technical ability judgment support system.

  A technical ability judgment support system that creates input information and information analyzed in the analysis result record database, and the creation of integrated scored information for each language and advice information of judgment results performed in the analysis part of the technical ability judgment support system The analysis result is displayed in the display part of.

  In the present embodiment, a process when the user requests evaluation of his or her own user skill will be described with reference to FIG.

  In S2010, the repository held by the user who has requested the evaluation of his or her skill is acquired from the open source information DB 1400. Here, the open source information DB 1400 refers to information obtained based on public data published on GitHub (registered trademark), BitBucket, GitLab, or the like.

  In step S2020, the repository is digitized (scored) based on the language type of the code (Perl, Python, etc.), the syntax analysis of the source code, and the number of issue / pull requests. Regarding the scoring method, for example, points / points to be added / deducted (for example, within a range of the maximum value of the points / deducted points, predetermined points are added / deducted according to the number of issues / pull requests, etc.) ) May be set in advance.

  In S2030, the relationship between the digitized repository and the deviation value is machine-learned. Here, the deviation value in the present embodiment indicates a relative skill level of a certain user in a set of users having programming skills. However, a certain number of users is required to indicate the relative skill level. For example, in the system of the present embodiment, a plurality of virtual users are set, and each virtual user is set. The score of the skill may be set in advance.

  In S2040, access is made to the DB in which the dependency relationship with another user such as the organization and the number of follow-ups is stored as a parameter, and the parameter is acquired.

  In S2050, the machine learning model is updated based on the acquired parameters.

  In S2060, the user skill is calculated as a deviation value. Here, as an additional process, the entire user's programming ability is calculated as a deviation value. Also, the deviation value in the individual language is calculated. Here, for the deviation value of each language, advice such as "You have strength in... If you want to raise the deviation value by 5 more times, increase the code XXX times every day." May be executed. In this analysis, the information is compared with the information already analyzed for each language, and a score for each language of the corresponding user is calculated. The analysis may provide advice on what to do to increase the score. Further, a process of visualizing the analyzed scores and advice for each language, including data not analyzed by the database, as a service on the web may be executed.

  In this embodiment, another mode of the second embodiment will be described.

  First, in this analysis, three pieces of quantization data relating to the user, quantization data relating to the repository, and quantization data relating to the pull request stored in the database relating to the relevant user are extracted.

  Next, a learning model generated by machine learning is prepared in advance, and parameters of each quantized data and its threshold value, reference value, and variance are defined.

  For the learning models and their defined data, information on each parameter is calculated from the extracted language-specific data, and a score in each language is calculated.

  The calculated score is converted into a deviation value as a skill deviation value from a predetermined reference value and variance.

Take an example.
(1) Assume that the total number of stars, which is data acquired from an API of a certain user, is 100 and the number of repositories is 20. (Examples when there are many other items, but analysis using two indices)
(2) The standardized parameter related to the total number of stars is set to a range 50 and an average value 30, and the standardized parameter related to the number of repositories is set to a range 40 and an average value 20 by machine learning in advance.
(3) For the machine learning model already created by support vector regression, a total skill deviation value is predicted using the data created in 1 and the data created in 2.
(4) From the total skill deviation value calculated by the prediction, the ratio of each language in the repository is calculated, and the skill deviation value in each language is calculated.

Here is another example.

  As a premise, the user A has information that is disclosed in the data source GitHub. Ruby, PHP, and Python programs are held and published in three types of repositories, respectively. Also, a Ruby issue is issued for an external repository that the user does not have. It is also assumed that a Ruby program is written as a pull request, a correction proposal is made, and the capture is completed.

Primary analysis processing 1 before database storage
At this time, since Ruby has issued an issue, it is recognized that the problem report information is being provided as an issue, and the Ruby has information on which repository the information has been issued. In addition, since a pull request is made to propose a modification of the Ruby program, and the corresponding repository has been completely imported, quantization is performed based on the following information. At this time, the number of issues, the number of pull requests, and the like, which are not provided by the API, are totaled, quantized, and stored in the database.

Primary analysis processing 2 before database storage
The user A also knows what organization he or she is in, and acquires the source code provided by the organization and the information of the user to which it belongs, tallies it, quantizes it, and stores it in the database.

Primary analysis processing 3 before database storage
In addition, there is information on what kind of data the repository has in the first place, evaluations by others for each repository, and information on what to do with and what people are following the repository, and based on that information, Based on the user information and the source code written by the user, the data is totaled and stored in a database in a quantized state.

Main analysis after storing in database Perform main analysis based on information stored in the database. In this analysis, as in one example, machine learning is performed by applying pre-analyzed and defined parameters such as range, average value, and variance to quantized data and each item to the machine learning model. Is calculated from the model. Based on the calculated data, the data is converted into a deviation value and provided to the user.

  In the present embodiment, the process when the user requests the evaluation of his or her own user skill and makes the evaluation request again after the time when the deviation value is first presented to the user (reference) will be described with reference to FIG. explain.

  In S3010, the time (time) when the deviation value is first calculated is stored. In this embodiment, this time is referred to as a reference time.

  In S3020, the information of the parameter at the reference time is obtained and stored.

  In S3030, the information of the parameter from the reference time to the current time is obtained and stored together with the parameter update time. Here, the current time point is a time point when the user newly requests an evaluation.

  In S3040, the weighting of each parameter is determined using a weighting curve graph as shown in FIG. The graph shown in FIG. 3B is configured such that the weight of the parameter updated at the time before the reference time increases, while the weight of the parameter updated at the time after the reference time decreases. I have.

  In S3050, a machine learning model (neural network, SVR (Support Vector Regression: support vector regression), etc.) is updated based on the newly updated parameters.

  In S3060, the user skill is calculated by a numerical value (deviation value). As in S2060, advice may be provided to the user.

  According to the present embodiment, even if an activity in which the user's own evaluation fluctuates is intentionally performed after the user's own evaluation, the fluctuation range of the user's evaluation is performed and falls within the range. Can be maintained.

  Although the embodiments of the present invention have been described above, various alternatives, modifications, or variations are possible for those skilled in the art based on the above description, and the present invention is not limited to the above various embodiments without departing from the spirit thereof. It is intended to cover alternatives, modifications or variations.

Technical ability judgment support system 1000, input means 1010, input information analysis means 1020, input comparison analysis means 1030, score calculation means 1040, input information or analysis result display means 1050, input information evaluation rule database 1110, input comparison analysis rules Database 1120, score calculation database 1130, input information and analysis result storage database 1140, control means 1200, user terminal 1300, open source information DB 1400

Claims (2)

On the computer,
A function to retrieve data about the repository held by a user from the database,
A function of digitizing the data relating to the repository based on at least a result of code parsing;
A function of storing a parameter related to data relating to the repository, wherein the parameter indicates a dependency between the user and another user,
A function for learning the relationship between the numerical value and the deviation value of the skill of the user by machine learning,
A function of updating a weighting coefficient between each layer of the learned neural network based on the parameter,
A function of calculating the skill deviation value of the user using the learned neural network;
The program to realize. The program according to claim 1,
The function of updating the weighting coefficient is based on the point in time when the skill deviation value is calculated, and relatively weights parameters in the past from the point in time, and reduces weight in parameters in the future from the point in time. And updating the weighting.