WO2024062964A1

WO2024062964A1 - Project execution assistance device, method, and program

Info

Publication number: WO2024062964A1
Application number: PCT/JP2023/033041
Authority: WO
Inventors: 九月貞光
Original assignee: 株式会社Vaiable
Priority date: 2022-09-22
Filing date: 2023-09-11
Publication date: 2024-03-28

Abstract

This project execution assistance device comprises: an estimation unit to which information about a project being formulated in a community is input, and which, using a model that outputs an action item or agenda required for executing the project, estimates an action item or agenda required for making decisions in the community; and an output unit that outputs the required action item or agenda that has been estimated.

Description

Project execution support devices, methods, and programs

The technology of the present disclosure relates to a project execution support device, method, and program.

Nowadays, DAOs (decentralized autonomous organizations) using blockchain clearly record the history of decision-making. Here, the DAO is an organization that is represented by rules defined by a computer program, such as a smart contract, and is controlled by holders of tokens necessary for decision-making (Patent Document 1).

In addition, through a programmed contract called a smart contract, rewards are automatically paid for the work performed at the DAO, for example.

People who have the role of AMA (ask me anything) play a major role in resolving the community's questions and promoting and revitalizing the community.

Japanese Patent Application Publication No. 2021-177422

The disclosed technology provides a project execution support device, method, and method that can support efficient community management using a model learned using data recording the decision-making history of projects planned in the community. The purpose is to provide programs.

A first aspect of the present disclosure is a project execution support device, which uses a model that inputs information about a project planned in a community and outputs action items or agenda items necessary for carrying out the project to generate ideas in the community. The device includes an estimating unit that estimates an action item or agenda item required for a decision, and an output unit that outputs the estimated necessary action item or agenda item.

A second aspect of the present disclosure is a project execution support device, which inputs an action item or agenda necessary for executing a project planned in a community and project information, and in which the action item or a generation unit that generates a program that rewards members who complete the action item or agenda in a community using a model that generates a program that rewards members who complete the agenda; and the generated program. an output unit that outputs the output.

A third aspect of the present disclosure is a project execution support device, which receives as input a program that rewards members who have completed the execution of action items or agendas necessary for the execution of a project planned in a community, and an explanatory text; an estimator that estimates a match rate between the program in the community and the input explanatory text using a model that estimates a match rate between the program and the explanatory text; and an estimator that outputs the estimated match rate. and an output section.

A fourth aspect of the present disclosure is a project execution support device, which inputs action items or agenda items necessary for executing a project planned in a community, and background information or desired remuneration of each member, and selects necessary action items. or an estimating unit that estimates members recommended as the person in charge of an action item or topic necessary in the community using a model that estimates members recommended as the person in charge of the topic; and an output that outputs the estimated members. including.

A fifth aspect of the present disclosure is a project execution support method, which uses a model that inputs information about a project planned in a community and outputs action items or agenda items necessary for carrying out the project, to generate ideas in the community. An action item or agenda item required for the decision is estimated, and the estimated necessary action item or agenda item is output.

A sixth aspect of the present disclosure is a project execution support method, which inputs an action item or agenda necessary for executing a project planned in a community and project information, and in which the action item or the project information is input. Using a model that generates a program that rewards members who complete the agenda, generate a program that rewards members who complete the action item or agenda in the community, and output the generated program. .

A seventh aspect of the present disclosure is a project execution support method, which receives as input a program that rewards members who have completed the execution of action items or agenda items necessary for the execution of a project planned in a community, and an explanatory text; Using a model for estimating the match rate between the program and the explanatory text, the match rate between the program in the community and the input explanatory text is estimated, and the estimated match rate is output.

An eighth aspect of the present disclosure is a project execution support method, which inputs action items or agenda items necessary for executing a project planned in a community and background information or desired remuneration of each member, and selects necessary action items. Alternatively, a model for estimating members recommended as persons in charge of an agenda item is used to estimate members recommended as persons in charge of an action item or topic required in the community, and the estimated members are output.

A ninth aspect of the present disclosure is a project execution support program, which is a program for causing a computer to function as the above-mentioned project execution support device.

According to the disclosed technology, efficient community management can be supported using a model learned using data that records the decision-making history of a project planned by a community.

FIG. 1 is a block diagram showing the configuration of an information processing system according to the present embodiment. FIG. 2 is a schematic block diagram of an example of a computer that functions as a management server of the present embodiment. FIG. 2 is a block diagram showing the configuration of a management server according to a first embodiment, a third embodiment, and a fourth embodiment. It is a flowchart which shows the content of the learning process routine of the management server based on this embodiment. It is a flowchart which shows the content of the estimation processing routine of the management server based on 1st Embodiment. FIG. 2 is a block diagram showing the configuration of a management server according to a second embodiment. It is a flowchart which shows the content of the estimation processing routine of the management server based on 2nd Embodiment. 12 is a flowchart showing the contents of an estimation processing routine of a management server according to a third embodiment. It is a flowchart which shows the content of the estimation processing routine of the management server based on 4th Embodiment.

Hereinafter, embodiments of the technology of the present disclosure will be described in detail with reference to the drawings.

[First embodiment]
As shown in Fig. 1, the information processing system 100 according to the first embodiment includes a management server 10 installed on the service management company side, and a user terminal 24 operated by a user. The management server 10 is an example of a project execution support device. For simplicity, Fig. 1 shows an example in which two user terminals 24 are provided, but three or more user terminals 24 may be provided.

The management server 10 and user terminals 24 are connected via a network 26 such as the Internet.

The user terminal 24 is a smartphone terminal, a mobile phone, a PDA (Personal Digital Assistant) terminal, a notebook/book type computer terminal, or the like.

FIG. 2 is a block diagram showing the hardware configuration of the management server 10 of this embodiment.

As shown in FIG. 2, the management server 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage 14, an input unit 15, a display unit 16, and a communication interface. interface ( I/F) 17. Each configuration is communicably connected to each other via a bus 19.

The CPU 11 is a central processing unit that executes various programs and controls various parts. That is, the CPU 11 reads a program from the ROM 12 or the storage 14 and executes the program using the RAM 13 as a work area. The CPU 11 controls each of the above components and performs various arithmetic operations according to programs stored in the ROM 12 or the storage 14. In this embodiment, the ROM 12 or the storage 14 stores programs for performing various processes.

The ROM 12 stores various programs and various data. The RAM 13 temporarily stores programs or data as a work area. The storage 14 is configured with an HDD (Hard Disk Drive) or an SSD (Solid State Drive), and stores various programs including an operating system and various data.

The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used to perform various input operations.

The display unit 16 is, for example, a liquid crystal display, and displays various information. The display section 16 may adopt a touch panel method and function as the input section 15.

The communication interface 17 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark), for example.

Next, the functional configuration of the management server 10 will be explained. As shown in FIG. 3, the management server 10 functionally includes a learning data generation section 30, a learning section 32, an acquisition section 34, an estimation section 36, an output section 38, a learning database (DB) 50, and a model storage section. 52, and a project information database (DB) 54.

The learning data generation unit 30 acquires the type of project from the project outline for an existing project planned by the community from the project information database 54.

Furthermore, the learning data generation unit 30 acquires "action items and agenda items" at each point in time for each "progress status of the project" for the existing project from the project information database 54, from the decision-making history of the project.

Here, if the "purpose" is to set up the next meeting, the "project type" is, for example, "meeting settings". The "progress status of a project" is, for example, the degree of accomplishment (percentage) associated with the consumption of action items to achieve the purpose of the project, the last action item consumed, or the degree of activity (message volume, etc.). This achievement level may be entered manually, the achievement level may be automatically calculated from the message content, or a program may be embedded to automatically monitor the achievement level. . Further, the "action items" are, for example, "determine attendees", "determine meeting contents", "adjust schedule", and "determine date".

Here, there are two possible ways to obtain the abstract concepts of project progress such as "attendee determination" and "schedule adjustment." Both methods are expected to improve accuracy when used in conjunction with the chronology of discussions within the community.

One is a method of interpreting natural language being discussed within a community and estimating clusters having identity. There are various ways to describe "schedule adjustment," but by vectorizing (embedding) natural language, it emerges as an ever-present vector in discussions aimed at "meeting settings." The timing of its appearance is often after "attendee determination". Based on these statistical trends, "schedule adjustment" is presented as the next action after "attendee determination". For embed here, it is possible to use a neural network or a large-scale language model (LLM) that has been trained in advance (Non-Patent Document 1).
[Non-patent Document 1]: GPT4 Internet Search <URL: https://openai.com/research/gpt-4.

Regarding project types, the types in the learning data are similarly clustered by vectorizing the natural language, and when using a trained model, it can be applied even if a new "project type" is input.

Another method is to associate each action with labels and tools that have already been assigned.

In the above example, "schedule adjustment" uses a schedule adjustment tool, and events that prompt community members to take action often occur. Furthermore, some community management tools have a function that allows labels to be prepared in advance and assigned to each task (Non-Patent Document 2).

By referring to the tools associated with each action and the setting information within the tools, "schedule adjustment" is presented as the next action after "attendee determination" as a frequency trend.
[Non-patent Document 2]: Development management tool redmine, Internet search <URL: https://redmine.jp>

Note that if no progress status is entered, a list of all actions required for "Project Type" will be output.

The learning data generation unit 30 generates a combination of “project type,” “project progress,” and “action item, agenda” as learning data, and stores it in the learning database 50.

For example, the combination of "meeting settings" as "project type", "attendee determination" as "project progress", and "schedule adjustment" as the next action "action item" can be used as learning data. generate.

The learning unit 32 learns a model for action estimation based on multiple pieces of learning data. The first input to this model is a vector indicating the "project type", and a vector indicating the "project progress" may also be input. The output of the model for action estimation is a vector indicating the "action item, agenda". The model for action estimation is, for example, a neural network.

The learning unit 32 stores the learned model for action estimation in the model storage unit 52.

The acquisition unit 34 acquires the type of project from the project summary for the project to be estimated planned by the community from the project information database 54.

Furthermore, the acquisition unit 34 acquires the "progress status of the project" from the project information database 54, regarding the project to be estimated, from the decision-making history of the project.

The estimation unit 36 inputs information about a project planned in the community and uses an action estimation model that outputs action items or agenda items necessary for carrying out the project to determine action items or agenda items necessary for decision making in the community. Estimate.

Here, the agenda is a topic that has not yet reached consensus in the community, for example, "What should be developed next?" Furthermore, an action item is something that has been agreed upon through discussion and turned into a task, and is, for example, a development task with a reward.

Specifically, the estimation unit 36 inputs a vector indicating the "type of project" and a vector indicating the "progress status of the project" to the learned model. The estimating unit 36 acquires the "action item, topic" required at the present time from the vector indicating "action item, topic" which is the model output.

The output unit 38 outputs the estimated necessary action items or topics.

The project information database 54 stores a project outline and a project decision history for each existing project. Furthermore, the project information database 54 stores a project outline and a project decision-making history regarding the project to be estimated. Here, the decision-making history is, for example, data representing discussions in natural language in a chat tool, or voting results based on tokens in decision-making.

Next, the operation of the information processing system 100 according to the first embodiment will be explained.

First, the management server 10 receives from each user terminal 24 a project outline and a project decision-making history as information regarding existing projects. The management server 10 registers the project outline and project decision history for each existing project in the project information database 54.

Then, upon receiving the learning instruction from the user terminal 24, the management server 10 executes the learning processing routine shown in FIG.

In step S100, the learning data generation unit 30 acquires the type of project from the project summary for each existing project planned by the community from the project information database 54.

In step S102, the learning unit 32 learns a model for action estimation based on multiple pieces of learning data, stores the learned model for action estimation in the model storage unit 52, and ends the learning processing routine.

Next, the management server 10 receives from the user terminal 24 the outline of the project and the decision-making history of the project as information regarding the project to be estimated. The management server 10 registers in the project information database 54 the project outline and project decision-making history for the project to be estimated.

Then, upon receiving the estimation instruction, the management server 10 executes the estimation processing routine shown in FIG. 5.

In step S110, the acquisition unit 34 acquires the type of project from the project summary from the project information database 54 for the project to be estimated planned by the community.

In step S112, the estimation unit 36 inputs a vector indicating the "type of project" and a vector indicating the "progress status of the project" to the learned model for action estimation. The estimating unit 36 acquires the "action item, topic" required at the present time from the vector indicating "action item, topic" which is the model output.

In step S114, the output unit 38 outputs the estimated necessary action items or topics.

As explained above, according to the information processing system according to the first embodiment, the information processing system according to the first embodiment uses a model that inputs information about a project planned by the community and outputs action items or agenda items necessary for carrying out the project. Estimate the action items or agenda needed for decision-making. In this way, efficient community management can be supported using a model learned using data that records the decision-making history of projects planned by the community.

In addition, by incorporating data that records decision-making history, such as DAO (for example, smart contract data), it is possible to learn models that support highly transparent and efficient community management. shall be.

Additionally, by inputting the "project type" and "project overview" when starting a project, features in the learning data that are similar to the project can be extracted and necessary action items and agenda items can be presented. In addition, when a project is in progress, by inputting the "progress status of the project" in addition to the above, it is possible to extract features in the learning data that are close to the current situation of the project and present necessary action items and agenda items.

[Second embodiment]
Next, an information processing system according to a second embodiment will be described. Note that parts having the same configuration as those in the first embodiment will be denoted by the same reference numerals and description thereof will be omitted.

The second embodiment differs from the first embodiment in that a model is used to generate a program that provides rewards to members who complete action items or agenda items in a community. This model takes as input action items or agenda items required for the execution of a project planned in the community and project information, and generates a program that provides rewards to members who complete action items or agenda items in the project information.

As shown in FIG. 6, the management server 210 according to the second embodiment includes a learning data generation section 230, a learning section 232, an acquisition section 234, a generation section 236, an output section 238, a learning database (DB) 250, and a model storage. 252 and a project information database (DB) 254.

The learning data generation unit 230 acquires the type of project from the project summary for an existing project planned by the community from the project information database 54.

Additionally, the learning data generation unit 230 acquires project decision-making history for existing projects from the project information database 54. The learning data generation unit 230 acquires "action items, agendas", "programs", and "final results of the project" at each point in time for each "progress status of the project" from the decision-making history of the project.

Here, the "program" is, for example, a program that has conditions and rewards determined by advance decision-making and automatically pays rewards when the conditions are met, and the chats and actions output by the members meet the conditions. Once met, rewards will be automatically paid to the member. An example of a program is a smart contract.

The learning data generation unit 230 generates information such as “type of project”, “progress status of project”,
A combination of "action item, agenda", "program", and "final result of the project" is generated as learning data and stored in the learning database 250.

Similarly to the learning unit 32 of the first embodiment, the learning unit 232 learns a model for action estimation, and stores the learned model for action estimation in the model storage unit 252.

The learning unit 232 learns a model for program generation based on a plurality of learning data. The first input to this model for program generation is a vector indicating the "type of project," and in addition, a vector indicating the "progress status of the project" may be input. The second input is a vector indicating "action item, agenda." The output of the model for program generation is a "reward granting program." The model is, for example, a neural network, and as an example, a modified version of seq2seq (text2code) (non-patent document 3), which inputs natural language and outputs source code, is used.

[Non-patent Document 3]: https://paperswithcode.com/paper/seq2sql-generating-structured-queries-from

This program generation model estimates an appropriate "reward" for the input and outputs the corresponding program.

Here, the "reward" is, for example, a token. Furthermore, the "final result of the project" includes, for example, the output (market value) of the community, the size of the community, and the like. Here, a specific example will be explained using a case where a reward is determined for an action "A". Assume that there are other past community histories α and β in which the action “A” was actually executed. It is also assumed that a reward A _α is given as a reward for the community history α, and a reward A _β is given for the community history β. Assume that the community history α has a result that the community size has increased, and the community history β has a result that the community size has not increased. In this case, the model is trained in such a way that the reward A _α is more correct as the reward for the action “A”.

The learning unit 232 stores the trained model for program generation in the model storage unit 252.

Note that a trained model for program generation may be obtained by reinforcement learning. In this case, each agent acts to maximize gain based on its own status (skill, etc.), and learns this behavior function from past learning data. The model estimates the "reward" that will be optimal for the final outcome of the project based on the current behavior function. Repeat this. By inputting input data to the finally learned reward estimation model, a program with appropriate rewards is output.

The acquisition unit 234 acquires the type of project from the project summary from the project information database 254 regarding the project to be estimated planned by the community.

Additionally, the acquisition unit 234 acquires the "progress status of the project" from the project decision-making history for the project to be estimated from the project information database 254.

Similarly to the estimating unit 36, the generating unit 236 estimates action items or agenda items necessary for decision-making in the community using a model for action estimation. The model for action estimation takes as input information about a project planned by the community and outputs action items or agenda items necessary for carrying out the project.

Specifically, similar to the estimation unit 36, the generation unit 236 inputs a vector indicating the “type of project” and a vector indicating the “progress status of the project” to the learned model for action estimation. The estimating unit 236 acquires the "action item, topic" required at the current moment from the vector indicating "action item, topic" which is the model output.

In addition, the generation unit 236 receives the estimated necessary action items or agenda items and project information as input, and uses the trained model for program generation to reward members who complete the action items or agenda items. Generate a program to grant.

Specifically, the generation unit 236 inputs a vector indicating "type of project," a vector indicating "progress status of project," and a vector indicating "action item, agenda" to the trained model generated by the program. The generation unit 236 obtains a source code indicating a "remuneration granting program" which is a model output.

In addition, a vector indicating "project type", a vector indicating "project progress status", and a vector indicating "action item, agenda" are input into the policy obtained by learning (supervised learning or reinforcement learning), Obtain the output "reward" and set it as a reward in the source code indicating the "reward granting program".

The output unit 238 outputs the generated program.

The project information database 254 stores project summaries and project decision-making histories for existing projects. Furthermore, the project information database 254 stores a project outline and a project decision-making history regarding the project to be estimated.

Next, the operation of the information processing system 100 according to the second embodiment will be explained.

First, the management server 10 receives from the user terminal 24 a project outline and a project decision-making history as information regarding an existing project. The management server 10 registers project summaries and project decision-making histories for existing projects in the project information database 54.

When the management server 10 receives a learning instruction from the user terminal 24, it executes a processing routine similar to the learning processing routine shown in FIG. 4 above.

Here, in step S100, the learning data generation unit 230 acquires the type of project from the project summary for an existing project planned by the community from the project information database 54.

In step S102, the learning unit 232 learns a model for action estimation based on a plurality of learning data, and stores the learned model for action estimation in the model storage unit 52. Then, the learning unit 232 learns a model for program generation based on the plurality of learning data, and stores the learned model for program generation in the model storage unit 52.

Then, the management server 10 receives from the user terminal 24 the outline of the project and the decision-making history of the project as information regarding the project to be estimated. The management server 10 registers in the project information database 54 the project outline and project decision-making history for the project to be estimated.

Then, upon receiving the estimation instruction, the management server 10 executes the estimation processing routine shown in FIG. 7.

In step S210, the acquisition unit 234 acquires the type of project from the project summary from the project information database 254 regarding the project to be estimated planned by the community.

In step S212, similarly to the estimation unit 36, the generation unit 236 inputs a vector indicating the “type of project” and a vector indicating the “progress status of the project” to the learned model for action estimation. The generation unit 236 acquires the "action item, agenda" required at the current time from the vector indicating the "action item, agenda" which is the model output.

In step S214, the generation unit 236 inputs a vector indicating "type of project," a vector indicating "progress status of project," and a vector indicating "action item, agenda" to the trained model generated by the program. The generation unit 236 obtains a source code indicating a "remuneration granting program" which is a model output.

In step S216, the generation unit 236 obtains a vector indicating "project type", a vector indicating "progress status of project", and a vector indicating "action item, agenda" through learning (supervised learning or reinforcement learning). input into the given policy, obtain the output "reward", and set it as a reward in the source code indicating the "reward granting program".

In step S218, the output unit 238 outputs the generated program.

As described above, according to the information processing system according to the second embodiment of the technology of the present disclosure, a model is used to generate a program that rewards members who complete the action item or agenda in the community. do. The model takes as input action items or agendas necessary for carrying out a project planned in the community and project information, and uses the project information to create a program that rewards members who complete the action items or agendas. generate. In this way, efficient community management can be supported using a model learned using data that records the decision-making history of projects planned by the community.

In addition, by incorporating data that records the process of decision-making, such as DAOs (for example, smart contract data), it will be possible to learn models that support transparent and efficient community management.

For example, when the "purpose" is to set up the next meeting, there is a smart contract in the past data that lists "determine attendees," "determine meeting content," "adjust schedule," and "determine date" as required action items. Suppose it exists. When trying to set up the same meeting in another DAO, it is thought that this smart contract can be used as is. For this reason, it is possible to generate a program corresponding to past data in which an action item and a smart contract are linked.

For example, when a community is first established, a DAO that makes games is equipped with a smart contract that allocates rewards to developers, and at some point, marketing becomes an issue and it becomes necessary to allocate rewards to marketers as well. . At this time, it is assumed that in a past similar project, the marketing department was allocated 30% of the remuneration compared to the development department, and as a result, the marketing measure was successful. In a project similar to this project, a program corresponding to the program of this project can be generated according to the progress status.

[Third embodiment]
Next, an information processing system according to a third embodiment will be described. Note that the configuration of the information processing system according to the third embodiment is the same as that of the first embodiment, so the same reference numerals are given and the explanation will be omitted.

The third embodiment differs from the first embodiment in that a model is used to estimate the match rate between the program in the community and the input explanatory text. The model inputs an explanatory text and a program that rewards members who complete action items or agendas necessary for carrying out a project planned by the community, and estimates the match rate between the program and the explanatory text.

The learning data generation unit 30 of the management server 10 according to the third embodiment obtains the project type from the project summary for existing projects planned by the community from the project information database 54.

The learning data generation unit 30 also obtains the project decision-making history for existing projects from the project information database 54. From the project decision-making history, the learning data generation unit 30 obtains the "program" and the "explanation given to the source code of the program" at each point in time for each "project progress status."

Here, the "explanatory text" is, for example, a specification or requirement definition document added during program development, or a requirement determined in natural language during DAO discussion.

The learning data generation unit 30 generates a combination of "project type", "project progress status", "correct program", and "explanatory text" as normal positive learning data, and stores it in the learning database 50. . Further, a combination of "project type", "project progress", "wrong program", and "explanation" is generated as normal negative learning data and stored in the learning database 50.

The learning unit 32 learns a model for correctness estimation based on a plurality of positive and negative learning data. The first input to this model for estimating correctness is a vector indicating "type of project", and in addition, a vector indicating "progress status of project" may be input.

The third input of the model for correctness estimation is text indicating "program source code", and the fourth input is text indicating "project explanatory text".

The output of the model for estimating accuracy is a scalar value indicating the "degree of accuracy." The model for correct/incorrect estimation is, for example, a neural network. As an example, a seq2seq (text2code) model can be used. Normally, a code suitable for the input is output, but in this embodiment, by calculating the generation probability of the source code that is input separately for the input text, if the probability is high, it is considered as a correct answer, and the probability If it is low, it is considered that there is a high possibility that it is wrong.

For example, for the action and reward pair "Pay 1 token to a user who has registered to participate",
if anyuser register calendar:
user Z += 1token
If 1 token is given to an unrelated user Z like this, it can be checked against the learned model and determined that there is a problem.

It is also possible to more specifically point out that there is a problem where "user Z" appears in the second line.

The learning unit 32 stores the trained model for correctness estimation in the model storage unit 52.

The acquisition unit 34 acquires the project type from the project overview of the estimated project planned by the community from the project information database 54.

Furthermore, the acquisition unit 34 acquires the "progress status of the project" from the project decision-making history for the project to be estimated from the project information database 54.

Furthermore, the acquisition unit 34 acquires, from the project information database 54, a source code indicating a "remuneration granting program" from the decision-making history of the project for the project to be estimated.

The acquisition unit 34 also acquires text input by the smart contract creator that indicates a description of the currently envisioned program.

The estimation unit 36 estimates the degree of accuracy using a model for estimating accuracy. The model for correct/incorrect estimation takes as input a program that rewards members who complete action items or agenda items necessary for carrying out a project planned in the community, and an explanatory text, and calculates the match rate between the program and the explanatory text. Estimate the degree of accuracy that represents.

Specifically, the estimating unit 36 inputs the source code indicating the "remuneration program" and the text of the input explanatory text into a model for correctness estimation, and obtains the "degree of correctness" that is the model output. do.

The output unit 38 outputs the estimated degree of accuracy.

The project information database 54 stores project summaries and project decision-making histories for existing projects. Furthermore, the project information database 54 stores a project outline and a project decision-making history regarding the project to be estimated.

Next, the operation of the information processing system 100 according to the third embodiment will be explained.

First, the management server 10 accepts information about an existing project, such as a project overview and a project decision-making history, from the user terminal 24. The management server 10 registers the project overview and project decision-making history for the existing project in the project information database 54.

Then, when the management server 10 receives a learning instruction from the user terminal 24, it executes a processing routine similar to the learning processing routine shown in Figure 4 above.

Here, in step S100, the learning data generation unit 30 acquires the type of project from the project summary for an existing project planned by the community from the project information database 54.

Furthermore, the learning data generation unit 30 acquires project decision-making history for existing projects from the project information database 54. The learning data generation unit 30 acquires the "program" and "explanatory text given to the source code of the program" at each point in time for each "progress status of the project" from the decision-making history of the project.

The learning data generation unit 30 generates a combination of "project type", "project progress status", "correct program", and "explanatory text" as normal positive learning data, and stores it in the learning database 50. . Further, a combination of "project type", "project progress status", "incorrect program", and "explanatory text" is generated as normal negative learning data and stored in the learning database 50.

In step S102, the learning unit 32 learns a model for correctness estimation based on a plurality of positive and negative learning data, and stores the learned model for correctness estimation in the model storage unit 52.

Then, upon receiving the estimation instruction, the management server 10 executes the estimation processing routine shown in FIG. 8.

In step S310, the acquisition unit 34 acquires the type of project from the project summary from the project information database 54 regarding the project to be estimated planned by the community.

Further, the acquisition unit 34 acquires, from the project information database 54, a source code indicating a "remuneration granting program" from the decision-making history of the project for the project to be estimated.

Additionally, the acquisition unit 34 acquires text input from the program creator that indicates the explanatory text of the program assumed at the present time.

In step S312, the estimation unit 36 inputs the source code indicating the "remuneration program" and the text of the input explanatory text into a model for correctness estimation, and obtains the "degree of correctness" that is the model output. .

In step S314, the output unit 38 outputs the estimated degree of correctness, and ends the estimation processing routine.

The program creator or community member checks the estimated "degree of correctness" and, if correct, adopts the obtained "program that gives rewards". On the other hand, if it is incorrect, the obtained program may not be appropriate, so check the obtained "reward granting program" and correct the source code.

As described above, according to the information processing system according to the third embodiment of the technology of the present disclosure, a model is used to estimate the match rate between the program in the community and the input explanatory text. The model inputs an explanatory text and a program that rewards members who complete action items or agenda items necessary for carrying out a project planned by the community, and estimates the match rate between the program and the explanatory text. In this way, efficient community management can be supported using a model learned using data that records the decision-making history of projects planned by the community.

For example, even if a malicious programmer creates a program for his own convenience, contrary to the rules established by the DAO, such malicious intent can be automatically detected.

In addition, an output model that takes a program as input and outputs a description of the program can be trained in advance, and the program can be input into the output model to determine the match rate between the output description and the input description.

[Fourth embodiment]
Next, an information processing system according to a fourth embodiment will be described. Note that the configuration of the information processing system according to the fourth embodiment is the same as that of the first embodiment, so the same reference numerals are given and the explanation will be omitted.

The fourth embodiment differs from the first embodiment in that a model is used to estimate members who are recommended as the person in charge of an action item or agenda item required in the community. The model takes as input the action items or agenda necessary to carry out the project planned by the community and the background information or desired reward of each member, and estimates the members recommended as the person in charge of the necessary action item or agenda. .

The learning data generation unit 30 of the management server 10 according to the fourth embodiment acquires the decision-making history of an existing project planned in a community from the project information database 54. The learning data generation unit 30 acquires, from the decision-making history of the project, "action items, agenda items,""members in charge," and "all members" including members not in charge.

Furthermore, the learning data generation unit 30 acquires "background information and desired remuneration" of "all members" from the member information for the existing project from the project information database 54. Here, the background information includes past project participation history or skills.

The learning data generation unit 30 generates a combination of "action items, agendas", "background information and desired remuneration" of "all members", and "members in charge" as learning data, and stores it in a learning database. 50.

Similarly to the learning unit 32 of the first embodiment, the learning unit 32 learns a model for action estimation, and stores the learned model for action estimation in the model storage unit 252.

The learning unit 32 learns a recommendation model based on a plurality of learning data. The first input of this model for recommendation is a vector indicating "action item, agenda", the second input is a vector indicating "background information" of each member, and in addition, each member You may input a vector indicating the "desired remuneration" of

The output of the recommendation model is a vector indicating the "recommendation degree" of each member. Models for recommendation include, for example, general recommendation algorithms (collaborative filtering), auction algorithms (real time bidding, etc.), neural networks (neural collaborative filtering), and the like.

Neural collaborative filtering is a method described in Non-Patent Document 4.
[Non-patent Document 4]: https://arxiv.org/pdf/1708.05031v2.pdf

In the specific process, learning progresses as follows. Note that in collaborative filtering and the like, the learning phase is not clearly separated, and new data is added to past data and then matrix operations are performed directly, but for convenience, inputting past data is referred to as the learning phase.

If there is no desired reward, create a matrix with members and actions as rows and columns for past data. This is background information. After determining the similarity between new actions and past actions, tasks are recommended to members. At this time, presentation/non-presentation may be switched in consideration of the amount of tasks the user is currently in charge of.

If there is a desired reward, and assuming the reward granting program described in the second embodiment above, the reward is determined in advance, so the task is presented only to members whose desired reward is reached. If the desired remuneration of any member has not been reached, the remuneration program described in the second embodiment is recalculated after adding the information "no takers" as the progress status. As a result, if there are cases in the past data where rewards have been increased in similar cases, the rewards can be increased accordingly.

The learning unit 32 stores the learned model for recommendation in the model storage unit 52.

The acquisition unit 34 also acquires the "project progress status" from the project decision-making history of the project to be estimated from the project information database 54.

Additionally, the acquisition unit 34 acquires each member's "background information" and "desired remuneration" for the project to be estimated from the project information database 54.

Similarly to the first embodiment, the generation unit 236 uses the action estimation model to estimate action items or agenda items necessary for decision-making in the community. The model for action estimation takes as input information about a project planned by the community and outputs action items or agenda items necessary for carrying out the project.

Specifically, similarly to the first embodiment, the generation unit 236 inputs a vector indicating the "type of project" and a vector indicating the "progress status of the project" to the learned model for action estimation. The generation unit 236 acquires the "action item, agenda" required at the current time from the vector indicating the "action item, agenda" which is the model output.

The estimation unit 36 inputs the action items or agenda necessary for carrying out the project planned in the community and the background information or desired reward of each member, and uses the learned model for recommendation to determine the necessary action items or agenda. Estimate the members recommended to take charge of the topic.

Specifically, the estimation unit 36 inputs a vector indicating "action item, agenda", "background information" of each member, and a vector indicating "desired reward" into a learned model for recommendation, and outputs the model. A vector indicating the "recommendation degree" of each member is obtained.

The output unit 38 outputs the estimated "recommendation degree" of each member.

Next, the operation of the information processing system 100 according to the fourth embodiment will be described.

First, the management server 10 receives information about an existing project, such as a project overview and a project decision-making history, from the user terminal 24. The management server 10 registers the project overview and project decision-making history for the existing project in the project information database 54.

Here, in step S100, the learning data generation unit 30 acquires the project decision-making history for an existing project planned by the community from the project information database 54. The learning data generation unit 30 acquires "action items, agenda items," "members in charge," and "all members," including members not in charge, from the decision-making history of the project.

Furthermore, the learning data generation unit 30 acquires "background information and desired remuneration" of "all members" from the member information for the existing project from the project information database 54.

The learning data generation unit 30 generates combinations of "action items, topics," "background information and desired remuneration" of "all members," and "members in charge" as learning data, and stores the data in the learning database 50.

In step S102, the learning unit 32 learns a model for action estimation based on a plurality of learning data, and stores the learned model for action estimation in the model storage unit 52. Then, the learning unit 32 learns a model for recommendation based on the plurality of learning data, and stores the learned model for recommendation in the model storage unit 52.

Then, upon receiving the estimation instruction, the management server 10 executes the estimation processing routine shown in FIG.

In step S410, the acquisition unit 34 acquires the type of project from the project summary from the project information database 54 for the project to be estimated planned by the community.

In step S412, the estimation unit 36 inputs a vector indicating the "type of project" and a vector indicating the "progress status of the project" to the learned model for action estimation. The estimating unit 236 acquires the "action item, topic" required at the current moment from the vector indicating "action item, topic" which is the model output.

In step S414, the acquisition unit 34 acquires each member's "background information" and "desired remuneration" for the project to be estimated from the project information database 54.

In step S416, the estimation unit 36 inputs a vector indicating "action item, agenda", "background information" of each member, and a vector indicating "desired reward" to the trained model for recommendation. The estimation unit 36 obtains a vector indicating the "recommendation degree" of each member, which is a model output.

In step S418, the output unit 38 outputs the estimated "recommendation level" of each member.

As described above, according to the information processing system according to the fourth embodiment of the technology of the present disclosure, a model is used to estimate members who are recommended as the person in charge of an action item or agenda required in the community. The model takes as input the action items or agenda necessary to carry out the project planned by the community and the background information or desired reward of each member, and estimates the members recommended as the person in charge of the necessary action item or agenda. . In this way, efficient community management can be supported using a model learned using data that records the decision-making history of projects planned by the community.

Note that the processing of the management server described above may be distributed and executed by multiple computers connected via a network.

Furthermore, the "action item" may include, for example, "watching an advertisement." In this case, the technology of the present disclosure may be applied to a mode in which an "advertisement" desired by an individual is distributed based on the individual's behavior record within the community, and the individual receives a reward by viewing the advertisement.

(Additional note)
Regarding the above embodiments, the following additional notes are further disclosed.

(Additional note 1)
memory and
at least one processor connected to the memory;
including;
The processor includes:
Estimate the action items or agenda necessary for decision-making in the community using a model that inputs information about a project planned in the community and outputs the action items or agenda necessary for carrying out the project,
A project execution support device that outputs the estimated necessary action items or agenda items.

(Additional note 2)
A non-temporary storage medium storing a program executable by a computer to execute project execution support processing,
The project execution support process is
Estimate the action items or agenda necessary for decision-making in the community using a model that inputs information about a project planned in the community and outputs the action items or agenda necessary for carrying out the project,
A non-temporary storage medium that outputs the estimated necessary action item or agenda.

(Additional note 3)
memory and
at least one processor connected to the memory;
including;
The processor includes:
Input the action items or agenda necessary for carrying out the project planned by the community and the project information,
Generate a program that rewards members who complete the action item or agenda in the community using a model that generates a program that rewards members who complete the action item or agenda using the project information. death,
A project execution support device that outputs the generated program.

(Additional note 4)
A non-temporary storage medium storing a program executable by a computer to execute project execution support processing,
The project execution support process is
Input the action items or agenda necessary for carrying out the project planned by the community and the project information,
Generate a program that rewards members who complete the action item or agenda in the community using a model that generates a program that rewards members who complete the action item or agenda using the project information. death,
A non-temporary storage medium that outputs the generated program.

(Additional Note 5)
Memory,
at least one processor coupled to the memory;
Including,
The processor,
A program for providing rewards to members who complete action items or agenda items necessary for the execution of a project planned in a community and an explanatory text are input, and a matching rate between the program and the explanatory text in the community is estimated using a model for estimating a matching rate between the program and the explanatory text;
The project execution support device outputs the estimated match rate.

(Additional note 6)
A non-temporary storage medium storing a program executable by a computer to execute project execution support processing,
The project execution support process is
A model that takes as input a program that rewards members who complete action items or agenda items necessary for carrying out a project planned by the community and an explanatory text, and estimates the match rate between the program and the explanatory text. to estimate the match rate between the program in the community and the input description,
A non-temporary storage medium that outputs the estimated matching rate.

(Additional Note 7)
Memory,
at least one processor coupled to the memory;
Including,
The processor,
Input the action items or agenda items required to carry out the project planned by the community, and each member's background information or desired compensation,
predicting members to be recommended as assignees for the required action items or topics in the community using a model for predicting members to be recommended as assignees for the required action items or topics;
The project execution support device outputs the estimated members.

(Supplementary Note 8)
A non-temporary storage medium storing a program executable by a computer to execute project execution support processing,
The project execution support process is
Input the action items or agenda necessary for carrying out the project planned by the community, and the background information or desired reward of each member.
Estimate the members recommended as the person in charge of the necessary action item or agenda in the community using a model for estimating the member recommended as the person in charge of the necessary action item or agenda,
A non-transitory storage medium that outputs the estimated members.

The disclosures of Japanese Application No. 2022-151940 and Japanese Application No. 2023-062413 are incorporated herein by reference in their entirety.

All documents, patent applications, and technical standards mentioned herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually indicated to be incorporated by reference. Incorporated herein by reference.

Claims

an estimation unit that estimates action items or agenda items necessary for decision-making in the community using a model that inputs information about a project planned in the community and outputs action items or agenda items necessary for carrying out the project;
an output unit that outputs the estimated necessary action item or agenda;
Project execution support equipment including.
Input the action items or agenda necessary for carrying out the project planned by the community and the project information,
Generate a program that rewards members who complete the action item or agenda in the community using a model that generates a program that rewards members who complete the action item or agenda using the project information. a generation unit to
an output unit that outputs the generated program;
Project execution support equipment including.
A model that takes as input a program that rewards members who complete action items or agenda items necessary for carrying out a project planned by the community and an explanatory text, and estimates the match rate between the program and the explanatory text. an estimating unit that estimates a match rate between the program in the community and the input explanatory text using the method;
an output unit that outputs the estimated matching rate;
Project execution support equipment including.
Input the action items or agenda necessary for carrying out the project planned by the community, and the background information or desired reward of each member.
an estimation unit that estimates members who are recommended as people in charge of a necessary action item or agenda in the community using a model that estimates members recommended as people in charge of the necessary action item or agenda;
an output unit that outputs the estimated members;
Project execution support equipment including.
The project execution support device according to claim 1, wherein the project information includes the content, type, or progress status of the project.
5. The project execution support device according to claim 4, wherein the background information includes past project participation history or skills.
Estimate the action items or agenda necessary for decision-making in the community using a model that inputs information about a project planned in the community and outputs the action items or agenda necessary for carrying out the project,
A project execution support method that outputs the estimated necessary action items or agenda items.
Input the action items or agenda necessary for carrying out the project planned by the community and the project information,
Generate a program that rewards members who complete the action item or agenda in the community using a model that generates a program that rewards members who complete the action item or agenda using the project information. death,
A project execution support method that outputs the generated program.
A model that takes as input a program that rewards members who complete action items or agenda items necessary for carrying out a project planned by the community and an explanatory text, and estimates the match rate between the program and the explanatory text. to estimate the match rate between the program in the community and the input description,
A project execution support method that outputs the estimated matching rate.
Input the action items or agenda necessary for carrying out the project planned by the community, and the background information or desired reward of each member.
Estimate the members recommended as the person in charge of the necessary action item or agenda in the community using a model for estimating the member recommended as the person in charge of the necessary action item or agenda,
A project execution support method that outputs the estimated members.
A project execution support program for causing a computer to function as the project execution support device according to any one of claims 1 to 6.