WO2021240832A1 - Processing device, processing method and processing program - Google Patents

Processing device, processing method and processing program Download PDF

Info

Publication number
WO2021240832A1
WO2021240832A1 PCT/JP2020/032016 JP2020032016W WO2021240832A1 WO 2021240832 A1 WO2021240832 A1 WO 2021240832A1 JP 2020032016 W JP2020032016 W JP 2020032016W WO 2021240832 A1 WO2021240832 A1 WO 2021240832A1
Authority
WO
WIPO (PCT)
Prior art keywords
evaluation
data
similarity
relation
evaluations
Prior art date
Application number
PCT/JP2020/032016
Other languages
French (fr)
Japanese (ja)
Inventor
真吾 小俣
Original Assignee
日本電信電話株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日本電信電話株式会社 filed Critical 日本電信電話株式会社
Priority to JP2022527476A priority Critical patent/JP7477791B2/en
Publication of WO2021240832A1 publication Critical patent/WO2021240832A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management

Definitions

  • the present invention relates to a processing apparatus, a processing method and a processing program.
  • the operation to improve the service is performed in the evaluation target of the person, organization, existing system, etc. related to the evaluation.
  • the work and analysis of analyzing the evaluation and associating the evaluation with the evaluation target is generally performed manually by a person.
  • Non-Patent Document 1 there is a method of expressing the development model of an AI (Artificial Intelligence) service system with ArchiMate (see Non-Patent Document 1). There is also a proposal to visualize the current situation and utilize it for selecting a system reconstruction method or supporting improvement points (see Non-Patent Document 2).
  • AI Artificial Intelligence
  • the present invention has been made in view of the above circumstances, and an object of the present invention is to provide a technique capable of easily associating an evaluation with an evaluation target.
  • the processing device has an acquisition unit for acquiring an evaluation for an evaluation target, a similarity calculation unit for calculating the similarity between each relation of an entity in the evaluation target, and an evaluation among the relations. It is provided with an output unit that outputs a relation having a high degree of similarity to.
  • the computer obtains an evaluation for the evaluation target, the computer calculates each relation of the entity in the evaluation target, and the computer calculates the similarity between the evaluations.
  • the computer calculates each relation of the entity in the evaluation target, and the computer calculates the similarity between the evaluations.
  • a step of outputting a relation having a high degree of similarity to the evaluation is provided.
  • One aspect of the present invention is a processing program that causes a computer to function as the processing device.
  • FIG. 1 is a diagram illustrating a functional block of a processing device according to an embodiment of the present invention.
  • FIG. 2 is a diagram illustrating a data structure of evaluation data and an example of the data.
  • FIG. 3 is a diagram illustrating a data structure of model data and an example of the data.
  • FIG. 4 is a diagram illustrating an example of the name of the entity used in the model data.
  • FIG. 5 is a diagram illustrating an example of a model configuration of an organization to be evaluated.
  • FIG. 6 is a diagram illustrating a data structure of similarity data and an example of the data.
  • FIG. 7 is a flowchart illustrating the acquisition process by the acquisition unit.
  • FIG. 8 is a flowchart illustrating the similarity calculation process by the similarity calculation unit.
  • FIG. 1 is a diagram illustrating a functional block of a processing device according to an embodiment of the present invention.
  • FIG. 2 is a diagram illustrating a data structure of evaluation data and an example of the data.
  • FIG. 9 is a diagram illustrating an example of an output result by the output unit.
  • FIG. 10 is a diagram illustrating a functional block of the output unit according to the first modification.
  • FIG. 11 is a diagram illustrating an example of a data structure of EA data.
  • FIG. 12 is a diagram illustrating a functional block of the output unit according to the second modification.
  • FIG. 13 is a diagram illustrating a learning unit according to the second modification.
  • FIG. 14 is a diagram illustrating a hardware configuration of a computer used in a processing device.
  • the processing apparatus 1 specifies an entity corresponding to the evaluation among the entities constituting the evaluation target organization 3 for the evaluation regarding the evaluation target organization 3 by computer processing.
  • the evaluation is a group of data in which the evaluation is meaningful for the organization to be evaluated, the products or services provided by the organization, etc. by the user, and is expressed in a language such as a sentence or a term.
  • the evaluation is one post of a comment by the user, one post of a word of mouth, and the like.
  • the evaluation data set may be customer complaint information, needs information, etc. managed by CRM (Customer Relationship Management) held by the evaluation target organization 3.
  • the processing device 1 includes evaluation data 11, model data 12, similarity data 13, acquisition unit 21, definition unit 22, similarity calculation unit 23, and output unit 24.
  • the evaluation data 11, the model data 12, and the similarity data 13 are data stored in the memory 902 or the storage 903.
  • the acquisition unit 21, the definition unit 22, the similarity calculation unit 23, and the output unit 24 are functional units mounted on the processing device 1 by executing a CPU 901 or a GPU (Graphics Processing Unit) (not shown).
  • the similarity calculation unit 23 may be executed by the GPU at the time of processing a neural network such as Word2vec or Doc2vec.
  • Evaluation data 11 is a set of evaluations. When the emotion score is calculated for each evaluation, the evaluation data 11 may associate the emotion score of the evaluation with each evaluation, for example, as shown in FIG.
  • the model data 12 is data that defines the relation of the entity related to the evaluation target organization 3. As shown in FIG. 3, the model data 12 associates the identifiers of the entities that define each relation. The identifier of an entity is associated with the name of the entity, as shown in FIG.
  • the operation model of the organization to be evaluated 3 is defined by a model language called ArchiMate.
  • the evaluation target organization 3 is divided into a plurality of entities.
  • the organization to be evaluated is divided into a plurality of layers, and one or more entities are associated with each layer.
  • the operation model of the evaluation target organization 3 is defined by the relation associated with the entities selected from each hierarchy.
  • the evaluation target organization 3 is divided into four layers, a tissue layer, an active structure layer, a behavior layer, and a passive structure layer, and one or more entities correspond to each layer. Be attached. Relations associate entities selected from each layer.
  • the model data 12 shown in FIG. 3 is formed corresponding to the model of the evaluation target organization 3 shown in FIG.
  • the model data 12 models the evaluation target organization 3 by the relation of each entity constituting the evaluation target organization 3.
  • the relation defined by the model data 12 is defined by a combination of the relation ID, the organization ID, the active structure ID, the behavior ID, and the passive structure ID.
  • the relation ID identifies the relation.
  • the organization ID, the active structure ID, the behavior ID, and the passive structure ID are IDs of entities selected from the organization layer, the active structure layer, the behavior layer, and the passive structure layer associated with each other in the relation. In the embodiment of the present invention, the IDs of the entities are numbered so as to be identifiable in each layer.
  • the organization ID "100,000”, the active structure ID "200,000”, the behavior ID "200,000”, and the passive structure ID “200,000” are associated with the relation ID "2" shown in FIG.
  • the organization ID "100,000” is a "customer support center” as shown in FIG. 4A in which the organization ID and the organization name are associated with each other.
  • the active structure ID “200,000” is a “sales representative” as shown in FIG. 4 (b) that associates the active structure ID with the active structure name.
  • the behavior ID "200,000” is “complaint correspondence” as shown in FIG. 4 (c) in which the behavior ID and the behavior name are associated with each other.
  • the passive structure ID "200,000” is “claim information” as shown in FIG. 4D in which the passive structure ID and the passive structure name are associated with each other.
  • the relation ID "2" corresponds to the "customer support center", “sales representative", “complaint handling” and “complaint information” in the evaluation target organization 3.
  • the similarity data 13 is data that associates the evaluation with the relationship between the evaluation target organization 3.
  • the similarity is calculated by the similarity calculation unit 23, which will be described later.
  • the similarity data 13 shown in FIG. 6 associates the relation ID with the highest similarity with the similarity between the relation IDs for each evaluation.
  • the acquisition unit 21 acquires the evaluation for the evaluation target.
  • the acquisition unit 21 acquires an evaluation regarding the evaluation target organization 3 from the evaluation providing device 2.
  • the evaluation providing device 2 is, for example, a posting site such as a microblog, an SNS (Social Networking Service) site, a word-of-mouth site, an in-house CRM system, or the like.
  • the acquisition unit 21 may acquire a plurality of evaluations.
  • the acquisition unit 21 may acquire the evaluation by the PULL type from the site or the sysstem, or may acquire the evaluation by the PUSH type.
  • the acquisition unit 21 may acquire an evaluation having a negative emotional score among the evaluations for the evaluation target.
  • the acquisition unit 21 may calculate an emotion score for each evaluation for the evaluation target and acquire a negative evaluation in which the emotion score is equal to or less than a predetermined value.
  • the higher the emotion score the more positive the impression of the evaluated tissue 3, and the lower the score, the negative the impression of the evaluated tissue 3.
  • the acquisition unit 21 calculates the emotion score of each evaluation by using, for example, the Google Natural Language API, and filters the evaluations having an emotion score of ⁇ 0.7 or less.
  • the acquisition unit 21 may acquire the evaluation corresponding to the name of the entity among the evaluations for the evaluation target.
  • the acquisition unit 21 may acquire an evaluation matching the search key by using, for example, the organization name, the active structure name, the behavior name, and the passive structure name of each table shown in FIG. 4 as the search key.
  • the search key may be the name of an entity corresponding to any structure among the organization layer, the active structure layer, the behavior layer, and the passive structure layer.
  • the search key may be the name of an entity that satisfies a predetermined condition.
  • the search key may be the name selected by the worker among the names of each entity.
  • the acquisition process by the acquisition unit 21 will be described with reference to FIG. 7.
  • the process shown in FIG. 7 is an example and is not limited to this.
  • step S101 the acquisition unit 21 collects evaluations that match the search word.
  • step S102 the acquisition unit 21 calculates an emotion score for each evaluation acquired in step S101.
  • step S103 the acquisition unit 21 filters the evaluation whose emotion score calculated in step S102 is equal to or less than the threshold value.
  • step S104 the acquisition unit 21 stores the evaluation filtered in step S103 in the evaluation data 11.
  • the definition unit 22 generates model data 12 for the evaluation target organization 3. For example, when an operation model is created by ArchiMate, the definition unit 22 acquires the output result and generates model data 12.
  • the similarity calculation unit 23 calculates the similarity between each relation of the entity of the operation model in the evaluation target and the evaluation.
  • the similarity between the relation and the evaluation is the similarity between the synthetic vector of the entity name of the relation and the synthetic vector of the words included in the evaluation.
  • the acquisition unit 21 acquires a plurality of evaluations
  • the similarity calculation unit 23 calculates the similarity between the evaluation and each relation for each of the plurality of evaluations.
  • the composite vector is calculated by, for example, python's gensim library and Word2Vec using a neural network.
  • the composite vector of the entity name of the relation is calculated.
  • the synthetic vector of the evaluation is calculated by processing the words included in the evaluation with the gensim library and Word2Vec. Here, words that frequently appear in the evaluation may be excluded.
  • the similarity between the relation synthesis vector and the evaluation synthesis vector is calculated by any method.
  • it may be Cos similarity, it may be calculated by Doc2Vec, or it may be calculated by a plurality of these calculation methods.
  • the similarity calculation unit 23 stores the similarity of each relation and each evaluation in the similarity data 13. Alternatively, as shown in FIG. 6, the similarity calculation unit 23 associates the evaluation with the ID of the relation having the highest similarity for the evaluation and stores it in the similarity data 13.
  • the similarity calculation process by the similarity calculation unit 23 will be described with reference to FIG.
  • the process shown in FIG. 8 is an example and is not limited to this.
  • step S201 the similarity calculation unit 23 converts each evaluation of the evaluation data 11 into a vector.
  • step S202 the similarity calculation unit 23 converts each relation of the model data 12 into a vector.
  • step S203 the similarity calculation unit 23 calculates the similarity between the vector of each evaluation calculated in step S201 and the vector of each relation calculated in step S202.
  • step S204 the similarity calculation unit 23 associates the evaluation with the relation ID having the highest similarity for each evaluation, and stores the evaluation in the similarity data 13.
  • the output unit 24 outputs a relation having a high degree of similarity to the evaluation among each relation. For example, the output unit 24 outputs a relation ID having the highest degree of similarity to the evaluation for each evaluation. The entity identified by the relation ID that has the highest similarity to the evaluation corresponds with reference to the evaluation.
  • the output unit 24 may output the number of evaluations having a similarity degree of a predetermined value or more for each relation. For example, as shown in FIG. 9, for each relation, the number of evaluations for which the relation is judged to have the highest degree of similarity is output. Since a relation with a large number of evaluations has many problems in improving the evaluation, it is possible to effectively reduce the negative evaluation by improving the relation with a large number of evaluations.
  • the output unit 24 outputs an index corresponding to the emotion score of the evaluation whose similarity is equal to or higher than a predetermined value for each relation. For example, the output unit 24 calculates and outputs an index having a positive correlation with the emotion score in the evaluation in which the relation is determined to have the highest degree of similarity for each relation.
  • the index is, for example, the average of the emotional scores of each evaluation, or the most negative emotional score of the emotional scores of each evaluation. Negative evaluations can be effectively reduced by improving with relations related to evaluations that strongly express negative emotions.
  • the processing device 1 can associate the evaluation with the evaluation target by computer processing, the cost for improving the evaluation can be significantly reduced. Further, since the relationship between the evaluation and the relationship in the evaluation target organization 3 can be indexed as the degree of similarity, the evaluation target organization 3 can efficiently work on the evaluation improvement.
  • the output unit 24a refers to the enterprise architecture data (EA data) 111 that has accumulated improvement results, acquires data related to the relation having a high degree of similarity to the evaluation from the EA data 111, and outputs the data.
  • the data output here can be an improvement measure in relations with high evaluation and similarity.
  • the output unit 24a includes an EA data 111, a recommendation data 112, a specific unit 121, a relation data 122, and a recommendation unit 123.
  • EA data 111 is enterprise architecture data.
  • the EA data 111 includes, for example, as shown in FIG. 11, the EA identifier, As-Is, To-Be, and Union-EA items.
  • An identifier that specifies the relationship between As-Is, To-Be, and Union-EA is set in the item of EA identifier.
  • the As-Is item the data of the As-Is model of the enterprise architecture model that has been improved by other companies is stored.
  • To-Be item the data of the To-Be model of the enterprise architecture model that has been improved by other companies is stored.
  • the Union-EA item the data of the integrated model of the enterprise architecture model that has been improved by other companies is stored.
  • the integrated model is a model in which the As-Is model, the To-Be model, and the Transition are linked. In the first modification, the case where the EA data 111 is generated in advance will be described.
  • the recommendation data 112 is data that specifies the improvement measures output by the recommendation unit 123.
  • the recommendation data 112 may be, for example, the EA identifier of the EA data 111, or may be the data of each item specified from the EA identifier.
  • the specific unit 121 outputs the relations having a high degree of similarity to the evaluation among the relations of the entities in the evaluation target as the relation data 122.
  • the relation data 122 may be a relation ID having the highest similarity with the evaluation as long as it can be converted into an entity name related to the relation, or each entity name converted from the relation ID having the highest similarity with the evaluation. May be.
  • the recommendation unit 123 refers to the EA data 111 and specifies the EA identifier related to the relation data 122.
  • the recommendation unit 123 searches the EA data 111 using the name of each entity specified from the relation data 122 as a search key, and identifies an EA identifier similar to the search key.
  • the recommendation unit 123 outputs the specified EA identifier or the data of each item specified from the EA identifier to the recommendation data 112.
  • the output unit 24a outputs the enterprise architecture model related to the relation for which improvement is required this time as a proposal for improvement measures from the past improvement measures by other companies and the like.
  • the improvement measures that solve the same problems as this time can be referred to.
  • the output unit 24a can output a relation highly related to the evaluation and can find a solution for the relation at an early stage.
  • the output unit 24b sets the As-Is model data of the EA data (enterprise architecture data) 111 as the input data of the encoder, and predicts the data of the To-Be model of the enterprise architecture data 111 as the input of the decoder and the correct answer data.
  • the model data 113 is learned, the prediction model data 113 is referred to, the relation data having a high degree of similarity to the evaluation is set as the input data, and the To-Be model data is output.
  • the data of the To-Be model is considered to be the future of relations with high evaluation and similarity, and can be an improvement measure.
  • the output unit 24b includes an EA data 111, a recommendation data 112, a prediction model data 113, a specific unit 121, a recommendation unit 123a, and a learning unit 124.
  • the EA data 111, the recommendation data 112, the specific unit 121, and the relation data 122 are as in the first modification.
  • the learning unit 124 learns by inputting the EA data 111 and outputs the prediction model data 113.
  • the learning unit 124 constructs a Seq2Seq (sequence to sequence) model by using teacher coercion using the EA data 111 that has already been improved in business.
  • the data of the As-Is model of the EA data 111 is set as the input of the encoder unit.
  • the data of the To-Be model of the EA data 111 is set as the input of the decoder unit and the correct answer data.
  • the data of the As-Is model and the data of the To-Be model of the EA data 111 are described in, for example, XML (Extensible Markup Language) format.
  • the learning unit 124 tunes the optimization algorithm and the like, and generates the prediction model data 113 using a model in which the error is sufficiently converged.
  • the learning unit 124 learns a prediction model that predicts the To-Be model from the As-Is model of the EA data.
  • the learning unit 124 learns using, for example, Python's Keras framework, the loss value obtained as a result of the learning becomes an error. When the loss value is as close to 0 as possible, it is determined that the error has sufficiently converged.
  • the recommendation unit 123a transfers the prediction model data 113 to the relation of the relation data 122 as an As-Is model, specifically, the entity name of the relation having a high degree of similarity to the evaluation. Enter.
  • the recommendation unit 123a acquires and outputs a To-Be model as an improvement measure in the relation specified by the specific unit 121.
  • the output unit 24b refers to the prediction model data 113 that has learned the past improvement measures by other companies and the like, and improves the To-Be model related to the relation that is required to be improved this time. Output as a plan. As a result, the output unit 24b can output a relation highly related to the evaluation and can find a solution for the relation at an early stage.
  • the processing device 1 of the present embodiment described above includes, for example, a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), and a communication device 904.
  • a general purpose computer system including an input device 905 and an output device 906 is used.
  • each function of the processing device 1 is realized by executing the processing program loaded on the memory 902 by the CPU 901.
  • a GPU may be used in combination with the CPU 901.
  • the processing device 1 may be mounted on one computer or may be mounted on a plurality of computers. Further, the processing device 1 may be a virtual machine mounted on a computer.
  • the processing program of the processing device 1 can be stored in a computer-readable recording medium such as an HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), DVD (Digital Versatile Disc), or via a network. It can also be delivered.
  • a computer-readable recording medium such as an HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), DVD (Digital Versatile Disc), or via a network. It can also be delivered.
  • the present invention is not limited to the above embodiment, and many modifications can be made within the scope of the gist thereof.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Human Resources & Organizations (AREA)
  • Operations Research (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Data Mining & Analysis (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

This processing device 1 comprises: an acquisition unit 21 for acquiring an assessment of an assessment target; a similarity calculating unit 23 for calculating a similarity between the assessment and respective relations of entities in the assessment target; and an output unit 24 for outputting a relation having a high similarity to the assessment, among the respective relations.

Description

処理装置、処理方法および処理プログラムProcessing equipment, processing methods and processing programs
　本発明は、処理装置、処理方法および処理プログラムに関する。 The present invention relates to a processing apparatus, a processing method and a processing program.
　会社等の組織に対する評価から、その評価に関連する人、組織、既存システム等の評価対象において、サービスを向上する運用が行われている。評価を分析して、評価と評価対象を対応づける作業および分析は、人が手動で行うことが一般的である。 From the evaluation of the organization such as a company, the operation to improve the service is performed in the evaluation target of the person, organization, existing system, etc. related to the evaluation. The work and analysis of analyzing the evaluation and associating the evaluation with the evaluation target is generally performed manually by a person.
　一般的に、ＡＩ（Artificial Intelligence）サービスシステムの開発モデルをArchiMateで表現する方法がある（非特許文献１参照）。また現状を視覚化して、システムの再構築手法の選択または改善点の支援に活用する提案もある（非特許文献２参照）。 Generally, there is a method of expressing the development model of an AI (Artificial Intelligence) service system with ArchiMate (see Non-Patent Document 1). There is also a proposal to visualize the current situation and utilize it for selecting a system reconstruction method or supporting improvement points (see Non-Patent Document 2).
　しかしながらいずれの文献も、コンピュータ処理により、評価と評価対象を対応づけることについて言及がない。 However, neither document mentions associating the evaluation with the evaluation target by computer processing.
　本発明は、上記事情に鑑みてなされたものであり、本発明の目的は、評価と評価対象を容易に対応づけることが可能な技術を提供することである。 The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a technique capable of easily associating an evaluation with an evaluation target.
　本発明の一態様の処理装置は、評価対象に対する評価を取得する取得部と、評価対象におけるエンティティの各リレーションと、評価との類似度を算出する類似度算出部と、各リレーションのうち、評価との類似度が高いリレーションを出力する出力部を備える。 The processing device according to one aspect of the present invention has an acquisition unit for acquiring an evaluation for an evaluation target, a similarity calculation unit for calculating the similarity between each relation of an entity in the evaluation target, and an evaluation among the relations. It is provided with an output unit that outputs a relation having a high degree of similarity to.
　本発明の一態様の処理方法は、コンピュータが、評価対象に対する評価を取得するステップと、コンピュータが、評価対象におけるエンティティの各リレーションと、評価との類似度を算出するステップと、コンピュータが、各リレーションのうち、評価との類似度が高いリレーションを出力するステップを備える。 In the processing method of one aspect of the present invention, the computer obtains an evaluation for the evaluation target, the computer calculates each relation of the entity in the evaluation target, and the computer calculates the similarity between the evaluations. Among the relations, a step of outputting a relation having a high degree of similarity to the evaluation is provided.
　本発明の一態様は、上記処理装置として、コンピュータを機能させる処理プログラムである。 One aspect of the present invention is a processing program that causes a computer to function as the processing device.
　本発明によれば、評価と評価対象を容易に対応づけることが可能な技術を提供することができる。 According to the present invention, it is possible to provide a technique capable of easily associating an evaluation with an evaluation target.
図１は、本発明の実施の形態に係る処理装置の機能ブロックを説明する図である。FIG. 1 is a diagram illustrating a functional block of a processing device according to an embodiment of the present invention. 図２は、評価データのデータ構造とデータの一例を説明する図である。FIG. 2 is a diagram illustrating a data structure of evaluation data and an example of the data. 図３は、モデルデータのデータ構造とデータの一例を説明する図である。FIG. 3 is a diagram illustrating a data structure of model data and an example of the data. 図４は、モデルデータに用いられるエンティティの名称の一例を説明する図である。FIG. 4 is a diagram illustrating an example of the name of the entity used in the model data. 図５は、評価対象組織のモデル構成の一例を説明する図である。FIG. 5 is a diagram illustrating an example of a model configuration of an organization to be evaluated. 図６は、類似度データのデータ構造とデータの一例を説明する図である。FIG. 6 is a diagram illustrating a data structure of similarity data and an example of the data. 図７は、取得部による取得処理を説明するフローチャートである。FIG. 7 is a flowchart illustrating the acquisition process by the acquisition unit. 図８は、類似度算出部による類似度算出処理を説明するフローチャートである。FIG. 8 is a flowchart illustrating the similarity calculation process by the similarity calculation unit. 図９は、出力部による出力結果の一例を説明する図である。FIG. 9 is a diagram illustrating an example of an output result by the output unit. 図１０は、第１の変形例に係る出力部の機能ブロックを説明する図である。FIG. 10 is a diagram illustrating a functional block of the output unit according to the first modification. 図１１は、ＥＡデータのデータ構造の一例を説明する図である。FIG. 11 is a diagram illustrating an example of a data structure of EA data. 図１２は、第２の変形例に係る出力部の機能ブロックを説明する図である。FIG. 12 is a diagram illustrating a functional block of the output unit according to the second modification. 図１３は、第２の変形例に係る学習部を説明する図である。FIG. 13 is a diagram illustrating a learning unit according to the second modification. 図１４は、処理装置に用いられるコンピュータのハードウエア構成を説明する図である。FIG. 14 is a diagram illustrating a hardware configuration of a computer used in a processing device.
　以下、図面を参照して、本発明の実施形態を説明する。図面の記載において同一部分には同一符号を付し説明を省略する。 Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the description of the drawings, the same parts are designated by the same reference numerals and the description thereof will be omitted.
　（処理装置）
　図１に示す本発明の実施の形態に係る処理装置１は、コンピュータ処理により、評価対象組織３に関する評価について、評価対象組織３を構成するエンティティのうち、その評価に対応するエンティティを特定する。評価は、ユーザによる、評価対象の組織、組織が提供する商品または役務等に対する、評価の意味のあるひとかたまりのデータであって、文章または用語等の言語で表現される。評価は、ユーザによるコメントの１投稿、口コミの１投稿等である。また評価データセットは、評価対象組織３が保持するＣＲＭ（Customer Relationship Management）で管理するお客様のクレーム情報、ニーズ情報等であっても良い。 (Processing device)
The processing apparatus 1 according to the embodiment of the present invention shown in FIG. 1 specifies an entity corresponding to the evaluation among the entities constituting the evaluation target organization 3 for the evaluation regarding the evaluation target organization 3 by computer processing. The evaluation is a group of data in which the evaluation is meaningful for the organization to be evaluated, the products or services provided by the organization, etc. by the user, and is expressed in a language such as a sentence or a term. The evaluation is one post of a comment by the user, one post of a word of mouth, and the like. Further, the evaluation data set may be customer complaint information, needs information, etc. managed by CRM (Customer Relationship Management) held by the evaluation target organization 3.
　処理装置１は、評価データ１１、モデルデータ１２、類似度データ１３、取得部２１、定義部２２、類似度算出部２３および出力部２４を備える。評価データ１１、モデルデータ１２および類似度データ１３は、メモリ９０２またはストレージ９０３に記憶されるデータである。取得部２１、定義部２２、類似度算出部２３および出力部２４は、CPU９０１またはGPU（Graphics Processing Unit）（図示せず）の実行によって処理装置１に実装される機能部である。ここで、類似度算出部２３は、Word2vec, Doc2vec等のニューラルネットワークの処理時において、GPUで実行されても良い。 The processing device 1 includes evaluation data 11, model data 12, similarity data 13, acquisition unit 21, definition unit 22, similarity calculation unit 23, and output unit 24. The evaluation data 11, the model data 12, and the similarity data 13 are data stored in the memory 902 or the storage 903. The acquisition unit 21, the definition unit 22, the similarity calculation unit 23, and the output unit 24 are functional units mounted on the processing device 1 by executing a CPU 901 or a GPU (Graphics Processing Unit) (not shown). Here, the similarity calculation unit 23 may be executed by the GPU at the time of processing a neural network such as Word2vec or Doc2vec.
　評価データ１１は、評価の集合である。各評価に対して感情スコアが算出された場合、評価データ１１は、例えば図２に示すように、各評価に、その評価の感情スコアを対応づけても良い。 Evaluation data 11 is a set of evaluations. When the emotion score is calculated for each evaluation, the evaluation data 11 may associate the emotion score of the evaluation with each evaluation, for example, as shown in FIG.
　モデルデータ１２は、評価対象組織３に関するエンティティのリレーションを定義するデータである。図３に示すように、モデルデータ１２は、各リレーションを定義するエンティティの識別子を紐付ける。エンティティの識別子は、図４に示すように、エンティティの名称に対応づけられる。 The model data 12 is data that defines the relation of the entity related to the evaluation target organization 3. As shown in FIG. 3, the model data 12 associates the identifiers of the entities that define each relation. The identifier of an entity is associated with the name of the entity, as shown in FIG.
　本発明の実施の形態において、評価対象組織３のオペレーションモデルは、ArchiMateと呼ばれるモデル言語で、定義される。評価対象組織３は、複数のエンティティに区分される。また評価対象組織を複数の階層にわけ、各階層に１以上のエンティティが紐付けられる。評価対象組織３のオペレーションモデルは、各階層から選択されたエンティティを紐付けたリレーションで定義される。本発明の実施の形態において、評価対象組織３は、図５に示すように、組織層、アクティブ構造層、振る舞い層およびパッシブ構造層の４つの層に区分され、各層に１以上のエンティティが対応づけられる。リレーションは、各層から選択されたエンティティを対応づける。 In the embodiment of the present invention, the operation model of the organization to be evaluated 3 is defined by a model language called ArchiMate. The evaluation target organization 3 is divided into a plurality of entities. In addition, the organization to be evaluated is divided into a plurality of layers, and one or more entities are associated with each layer. The operation model of the evaluation target organization 3 is defined by the relation associated with the entities selected from each hierarchy. In the embodiment of the present invention, as shown in FIG. 5, the evaluation target organization 3 is divided into four layers, a tissue layer, an active structure layer, a behavior layer, and a passive structure layer, and one or more entities correspond to each layer. Be attached. Relations associate entities selected from each layer.
　図３に示すモデルデータ１２は、図５に示す評価対象組織３のモデルに対応して形成される。モデルデータ１２は、評価対象組織３を構成する各エンティティのリレーションにより、評価対象組織３をモデル化する。モデルデータ１２が定義するリレーションは、リレーションＩＤ、組織ＩＤ、アクティブ構造ＩＤ、振る舞いＩＤおよびパッシブ構造ＩＤの組み合わせにより定義される。リレーションＩＤは、リレーションを識別する。組織ＩＤ、アクティブ構造ＩＤ、振る舞いＩＤおよびパッシブ構造ＩＤは、それぞれ、そのリレーションで対応づけられる組織層、アクティブ構造層、振る舞い層およびパッシブ構造層から選択されたエンティティのＩＤである。なお本発明の実施の形態において、エンティティのＩＤは、各層において識別可能に採番される。 The model data 12 shown in FIG. 3 is formed corresponding to the model of the evaluation target organization 3 shown in FIG. The model data 12 models the evaluation target organization 3 by the relation of each entity constituting the evaluation target organization 3. The relation defined by the model data 12 is defined by a combination of the relation ID, the organization ID, the active structure ID, the behavior ID, and the passive structure ID. The relation ID identifies the relation. The organization ID, the active structure ID, the behavior ID, and the passive structure ID are IDs of entities selected from the organization layer, the active structure layer, the behavior layer, and the passive structure layer associated with each other in the relation. In the embodiment of the present invention, the IDs of the entities are numbered so as to be identifiable in each layer.
　図３に示すリレーションＩＤ「２」に、組織ＩＤ「１０００００」、アクティブ構造ＩＤ「２０００００」、振る舞いＩＤ「２０００００」およびパッシブ構造ＩＤ「２０００００」が対応づけられる。組織ＩＤ「１０００００」は、組織ＩＤと組織名を対応づける図４（ａ）が示すように、「カスタマーサポートセンタ」である。アクティブ構造ＩＤ「２０００００」は、アクティブ構造ＩＤとアクティブ構造名を対応づける図４（ｂ）が示すように、「営業担当」である。振る舞いＩＤ「２０００００」は、振る舞いＩＤと振る舞い名を対応づける図４（ｃ）が示すように、「クレーム対応」である。パッシブ構造ＩＤ「２０００００」は、パッシブ構造ＩＤとパッシブ構造名を対応づける図４（ｄ）が示すように、「クレーム情報」である。リレーションＩＤ「２」は、評価対象組織３における「カスタマーサポートセンタ」、「営業担当」、「クレーム対応」および「クレーム情報」に対応する。 The organization ID "100,000", the active structure ID "200,000", the behavior ID "200,000", and the passive structure ID "200,000" are associated with the relation ID "2" shown in FIG. The organization ID "100,000" is a "customer support center" as shown in FIG. 4A in which the organization ID and the organization name are associated with each other. The active structure ID “200,000” is a “sales representative” as shown in FIG. 4 (b) that associates the active structure ID with the active structure name. The behavior ID "200,000" is "complaint correspondence" as shown in FIG. 4 (c) in which the behavior ID and the behavior name are associated with each other. The passive structure ID "200,000" is "claim information" as shown in FIG. 4D in which the passive structure ID and the passive structure name are associated with each other. The relation ID "2" corresponds to the "customer support center", "sales representative", "complaint handling" and "complaint information" in the evaluation target organization 3.
　類似度データ１３は、評価と、評価対象組織３のリレーションとの類似度を対応づけるデータである。類似度は、後述の類似度算出部２３によって算出される。図６に示す類似度データ１３は、各評価について、最も類似度が高いリレーションＩＤと、そのリレーションＩＤとの類似度を対応づける。 The similarity data 13 is data that associates the evaluation with the relationship between the evaluation target organization 3. The similarity is calculated by the similarity calculation unit 23, which will be described later. The similarity data 13 shown in FIG. 6 associates the relation ID with the highest similarity with the similarity between the relation IDs for each evaluation.
　取得部２１は、評価対象に対する評価を取得する。取得部２１は、評価提供装置２から、評価対象組織３に関する評価を取得する。評価提供装置２は、例えば、マイクロブログ等の投稿サイト、ＳＮＳ（Social Networking Service）サイト、口コミサイト、自社のＣＲＭシステム等である。取得部２１は、複数の評価を取得しても良い。取得部２１は、サイトまたはシスステムからPULL型で評価を取得しても良いし、PUSH型で評価を取得しても良い。 The acquisition unit 21 acquires the evaluation for the evaluation target. The acquisition unit 21 acquires an evaluation regarding the evaluation target organization 3 from the evaluation providing device 2. The evaluation providing device 2 is, for example, a posting site such as a microblog, an SNS (Social Networking Service) site, a word-of-mouth site, an in-house CRM system, or the like. The acquisition unit 21 may acquire a plurality of evaluations. The acquisition unit 21 may acquire the evaluation by the PULL type from the site or the sysstem, or may acquire the evaluation by the PUSH type.
　取得部２１は、評価対象に対する評価のうち、感情スコアがネガティブな評価を取得しても良い。取得部２１は、評価対象に対する各評価について感情スコアを算出し、感情スコアが所定の値以下となるネガティブな評価を取得しても良い。本発明の実施の形態において感情スコアは、高いほど評価対象組織３に対するポジティブな印象を意味し、低いほど評価対象組織３に対するネガティブな印象を意味する。取得部２１は、例えば、Google Natural Language APIを用いて、各評価の感情スコアを算出し、感情スコアが－０．７以下の評価をフィルタリングする。 The acquisition unit 21 may acquire an evaluation having a negative emotional score among the evaluations for the evaluation target. The acquisition unit 21 may calculate an emotion score for each evaluation for the evaluation target and acquire a negative evaluation in which the emotion score is equal to or less than a predetermined value. In the embodiment of the present invention, the higher the emotion score, the more positive the impression of the evaluated tissue 3, and the lower the score, the negative the impression of the evaluated tissue 3. The acquisition unit 21 calculates the emotion score of each evaluation by using, for example, the Google Natural Language API, and filters the evaluations having an emotion score of −0.7 or less.
　取得部２１は、評価対象に対する評価のうち、エンティティの名称に対応する評価を取得しても良い。取得部２１は、例えば、図４に示す各テーブルの組織名、アクティブ構造名、振る舞い名およびパッシブ構造名を、検索キーとして、検索キーに一致する評価を取得しても良い。検索キーは、組織層、アクティブ構造層、振る舞い層およびパッシブ構造層のうち、いずれかの構造に対応するエンティティの名称であっても良い。検索キーは、所定条件を満たすエンティティの名称であっても良い。検索キーは、各エンティティの名称のうち、作業者によって選択された名称であっても良い。 The acquisition unit 21 may acquire the evaluation corresponding to the name of the entity among the evaluations for the evaluation target. The acquisition unit 21 may acquire an evaluation matching the search key by using, for example, the organization name, the active structure name, the behavior name, and the passive structure name of each table shown in FIG. 4 as the search key. The search key may be the name of an entity corresponding to any structure among the organization layer, the active structure layer, the behavior layer, and the passive structure layer. The search key may be the name of an entity that satisfies a predetermined condition. The search key may be the name selected by the worker among the names of each entity.
　図７を参照して、取得部２１による取得処理を説明する。なお図７に示す処理は一例であって、これに限るものではない。 The acquisition process by the acquisition unit 21 will be described with reference to FIG. 7. The process shown in FIG. 7 is an example and is not limited to this.
　まずステップＳ１０１において取得部２１は、検索ワードに適合する評価を収集する。ステップＳ１０２において取得部２１は、ステップＳ１０１で取得した各評価について、感情スコアを算出する。 First, in step S101, the acquisition unit 21 collects evaluations that match the search word. In step S102, the acquisition unit 21 calculates an emotion score for each evaluation acquired in step S101.
　ステップＳ１０３において取得部２１は、ステップＳ１０２で算出された感情スコアが閾値以下の評価をフィルタリングする。ステップＳ１０４において取得部２１は、ステップＳ１０３でフィルタリングされた評価を、評価データ１１に格納する。 In step S103, the acquisition unit 21 filters the evaluation whose emotion score calculated in step S102 is equal to or less than the threshold value. In step S104, the acquisition unit 21 stores the evaluation filtered in step S103 in the evaluation data 11.
　定義部２２は、評価対象組織３について、モデルデータ１２を生成する。例えば、ArchiMateでオペレーションモデルが作成された場合、定義部２２は、その出力結果を取得しして、モデルデータ１２を生成する。 The definition unit 22 generates model data 12 for the evaluation target organization 3. For example, when an operation model is created by ArchiMate, the definition unit 22 acquires the output result and generates model data 12.
　類似度算出部２３は、評価対象におけるオペレーションモデルのエンティティの各リレーションと、評価との類似度を算出する。リレーションと評価との類似度は、リレーションのエンティティ名の合成ベクトルと、評価に含まれる単語の合成ベクトルとの、類似度である。取得部２１が複数の評価を取得する場合、類似度算出部２３は、複数の評価のそれぞれについて、評価と各リレーションとの類似度を算出する。 The similarity calculation unit 23 calculates the similarity between each relation of the entity of the operation model in the evaluation target and the evaluation. The similarity between the relation and the evaluation is the similarity between the synthetic vector of the entity name of the relation and the synthetic vector of the words included in the evaluation. When the acquisition unit 21 acquires a plurality of evaluations, the similarity calculation unit 23 calculates the similarity between the evaluation and each relation for each of the plurality of evaluations.
　合成ベクトルは、例えば、pythonのgensimライブラリと、ニューラルネットワークを用いるWord2Vecにより算出される。リレーションのエンティティ名を、gensimライブラリとWord2Vecで処理することにより、リレーションのエンティティ名の合成ベクトルが算出される。また評価に含まれる単語を、gensimライブラリとWord2Vecで処理することにより、評価の合成ベクトルが算出される。ここで、評価のうち、頻出する単語は除外されても良い。 The composite vector is calculated by, for example, python's gensim library and Word2Vec using a neural network. By processing the entity name of the relation with the gensim library and Word2Vec, the composite vector of the entity name of the relation is calculated. In addition, the synthetic vector of the evaluation is calculated by processing the words included in the evaluation with the gensim library and Word2Vec. Here, words that frequently appear in the evaluation may be excluded.
　リレーションの合成ベクトルと評価の合成ベクトルの類似度は、任意の方法で算出される。例えば、Ｃｏｓ類似度であっても良いし、Ｄｏｃ２Ｖｅｃによって算出されても良いし、これらの複数の算出方法で算出されても良い。 The similarity between the relation synthesis vector and the evaluation synthesis vector is calculated by any method. For example, it may be Cos similarity, it may be calculated by Doc2Vec, or it may be calculated by a plurality of these calculation methods.
　類似度算出部２３は、各リレーションおよび各評価の類似度を、類似度データ１３に格納する。あるいは、類似度算出部２３は、図６に示すように、評価と、その評価について最も類似度が高いリレーションのＩＤを対応づけて、類似度データ１３に格納する。 The similarity calculation unit 23 stores the similarity of each relation and each evaluation in the similarity data 13. Alternatively, as shown in FIG. 6, the similarity calculation unit 23 associates the evaluation with the ID of the relation having the highest similarity for the evaluation and stores it in the similarity data 13.
　図８を参照して、類似度算出部２３による類似度算出処理を説明する。なお図８に示す処理は一例であって、これに限るものではない。 The similarity calculation process by the similarity calculation unit 23 will be described with reference to FIG. The process shown in FIG. 8 is an example and is not limited to this.
　ステップＳ２０１において類似度算出部２３は、評価データ１１の各評価をベクトルに変換する。ステップＳ２０２において類似度算出部２３は、モデルデータ１２の各リレーションをベクトルに変換する。 In step S201, the similarity calculation unit 23 converts each evaluation of the evaluation data 11 into a vector. In step S202, the similarity calculation unit 23 converts each relation of the model data 12 into a vector.
　ステップＳ２０３において類似度算出部２３は、ステップＳ２０１で算出した各評価のベクトルと、ステップＳ２０２で算出した各リレーションのベクトルとの類似度を算出する。ステップＳ２０４において類似度算出部２３は、各評価について、その評価と最も類似度が高いリレーションＩＤを対応づけて、類似度データ１３に格納する。 In step S203, the similarity calculation unit 23 calculates the similarity between the vector of each evaluation calculated in step S201 and the vector of each relation calculated in step S202. In step S204, the similarity calculation unit 23 associates the evaluation with the relation ID having the highest similarity for each evaluation, and stores the evaluation in the similarity data 13.
　出力部２４は、各リレーションのうち、評価との類似度が高いリレーションを出力する。例えば、出力部２４は、各評価について、その評価と最も類似度が高いリレーションＩＤを出力する。評価と最も類似度が高いリレーションＩＤで特定されるエンティティが、その評価を参照して対応する。 The output unit 24 outputs a relation having a high degree of similarity to the evaluation among each relation. For example, the output unit 24 outputs a relation ID having the highest degree of similarity to the evaluation for each evaluation. The entity identified by the relation ID that has the highest similarity to the evaluation corresponds with reference to the evaluation.
　出力部２４は、リレーション毎に、類似度が所定値以上の評価の数を出力しても良い。例えば図９に示すように、リレーション毎に、そのリレーションが最も類似度が高いと判断された評価の数を出力する。評価の数が多いリレーションは、評価改善における課題が多いことから、評価数が多いリレーションで改善することにより、効果的にネガティブ評価を減らすことができる。 The output unit 24 may output the number of evaluations having a similarity degree of a predetermined value or more for each relation. For example, as shown in FIG. 9, for each relation, the number of evaluations for which the relation is judged to have the highest degree of similarity is output. Since a relation with a large number of evaluations has many problems in improving the evaluation, it is possible to effectively reduce the negative evaluation by improving the relation with a large number of evaluations.
　出力部２４は、リレーション毎に、類似度が所定値以上の評価の感情スコアに対応する指標を出力する。出力部２４は、例えば、リレーション毎に、そのリレーションが最も類似度が高いと判断された評価における感情スコアと正の相関を有する指標を算出して出力する。指標は、例えば、各評価の感情スコアの平均、または各評価の感情スコアのうち最もネガティブな感情スコア等である。ネガティブな感情を強く表現した評価に関連するリレーションで改善することにより、効果的にネガティブ評価を減らすことができる。 The output unit 24 outputs an index corresponding to the emotion score of the evaluation whose similarity is equal to or higher than a predetermined value for each relation. For example, the output unit 24 calculates and outputs an index having a positive correlation with the emotion score in the evaluation in which the relation is determined to have the highest degree of similarity for each relation. The index is, for example, the average of the emotional scores of each evaluation, or the most negative emotional score of the emotional scores of each evaluation. Negative evaluations can be effectively reduced by improving with relations related to evaluations that strongly express negative emotions.
　本発明の実施の形態に係る処理装置１は、コンピュータ処理により、評価と評価対象を対応づけることができるので、評価改善におけるコストを大幅に削減することができる。また評価と、評価対象組織３におけるリレーションとの関係を類似度として指標化できるので、評価対象組織３は、効率的に評価改善に取り組むことができる。 Since the processing device 1 according to the embodiment of the present invention can associate the evaluation with the evaluation target by computer processing, the cost for improving the evaluation can be significantly reduced. Further, since the relationship between the evaluation and the relationship in the evaluation target organization 3 can be indexed as the degree of similarity, the evaluation target organization 3 can efficiently work on the evaluation improvement.
　本発明の実施の形態において、処理装置１の出力部２４が、評価と類似度の高いリレーションを出力する場合を説明した。第１の変形例および第２の変形例において、評価と類似度の高いリレーションに対する改善策を出力する場合を説明する。 In the embodiment of the present invention, the case where the output unit 24 of the processing device 1 outputs a relation having a high degree of similarity to the evaluation has been described. In the first modification and the second modification, a case of outputting an improvement measure for a relation having a high degree of evaluation and similarity will be described.
　（第１の変形例）
　図１０を参照して、第１の変形例に係る出力部２４ａを説明する。出力部２４ａは、改善実績を蓄積したエンタープライズアーキテクチャデータ（ＥＡデータ）１１１を参照して、ＥＡデータ１１１から、評価との類似度が高いリレーションに関連するデータを取得して出力する。ここで出力されるデータは、評価と類似度の高いリレーションにおける改善策となりうる。 (First modification)
The output unit 24a according to the first modification will be described with reference to FIG. 10. The output unit 24a refers to the enterprise architecture data (EA data) 111 that has accumulated improvement results, acquires data related to the relation having a high degree of similarity to the evaluation from the EA data 111, and outputs the data. The data output here can be an improvement measure in relations with high evaluation and similarity.
　出力部２４ａは、ＥＡデータ１１１、レコメンドデータ１１２、特定部１２１、リレーションデータ１２２およびレコメンド部１２３を備える。 The output unit 24a includes an EA data 111, a recommendation data 112, a specific unit 121, a relation data 122, and a recommendation unit 123.
　ＥＡデータ１１１は、エンタープライズアーキテクチャデータである。ＥＡデータ１１１は、例えば図１１に示すように、ＥＡ識別子、Ａｓ－Ｉｓ、Ｔｏ－ＢｅおよびＵｎｉｏｎ－ＥＡの各項目を備える。ＥＡ識別子の項目に、Ａｓ－Ｉｓ、Ｔｏ－ＢｅおよびＵｎｉｏｎ－ＥＡの関係を特定する識別子が設定される。Ａｓ－Ｉｓの項目に、他の企業などが改善を実現したエンタープライズアーキテクチャモデルのＡｓ－Ｉｓモデルのデータが格納される。Ｔｏ－Ｂｅの項目に、他の企業などが改善を実現したエンタープライズアーキテクチャモデルのＴｏ－Ｂｅモデルのデータが格納される。Ｕｎｉｏｎ－ＥＡの項目に、他の企業などが改善を実現したエンタープライズアーキテクチャモデルの統合モデルのデータが格納される。統合モデルは、Ａｓ－Ｉｓモデル、Ｔｏ－ＢｅモデルおよびＴｒａｎｓｉｔｉｏｎが連結したモデルである。第１の変形例において、ＥＡデータ１１１は、予め生成されている場合を説明する。 EA data 111 is enterprise architecture data. The EA data 111 includes, for example, as shown in FIG. 11, the EA identifier, As-Is, To-Be, and Union-EA items. An identifier that specifies the relationship between As-Is, To-Be, and Union-EA is set in the item of EA identifier. In the As-Is item, the data of the As-Is model of the enterprise architecture model that has been improved by other companies is stored. In the To-Be item, the data of the To-Be model of the enterprise architecture model that has been improved by other companies is stored. In the Union-EA item, the data of the integrated model of the enterprise architecture model that has been improved by other companies is stored. The integrated model is a model in which the As-Is model, the To-Be model, and the Transition are linked. In the first modification, the case where the EA data 111 is generated in advance will be described.
　レコメンドデータ１１２は、レコメンド部１２３が出力する改善策を特定するデータである。レコメンドデータ１１２は、例えばＥＡデータ１１１のＥＡ識別子であっても良いし、ＥＡ識別子から特定される各項目のデータであっても良い。 The recommendation data 112 is data that specifies the improvement measures output by the recommendation unit 123. The recommendation data 112 may be, for example, the EA identifier of the EA data 111, or may be the data of each item specified from the EA identifier.
　特定部１２１は、本発明の実施の形態にかかる出力部２４の処理と同様で、評価対象におけるエンティティの各リレーションのうち、評価との類似度が高いリレーションを、リレーションデータ１２２として出力する。リレーションデータ１２２は、リレーションに関連するエンティティ名に変換できればよく、評価と最も類似度が高いリレーションＩＤであっても良いし、評価と最も類似度が高いリレーションＩＤから変換された各エンティティ名であっても良い。 Similar to the processing of the output unit 24 according to the embodiment of the present invention, the specific unit 121 outputs the relations having a high degree of similarity to the evaluation among the relations of the entities in the evaluation target as the relation data 122. The relation data 122 may be a relation ID having the highest similarity with the evaluation as long as it can be converted into an entity name related to the relation, or each entity name converted from the relation ID having the highest similarity with the evaluation. May be.
　レコメンド部１２３は、ＥＡデータ１１１を参照して、リレーションデータ１２２に関連するＥＡ識別子を特定する。レコメンド部１２３は、リレーションデータ１２２から特定される各エンティティの名称を検索キーとして、ＥＡデータ１１１を検索し、検索キーに類似するＥＡ識別子を特定する。レコメンド部１２３は、特定されたＥＡ識別子、またはＥＡ識別子から特定される各項目のデータを、レコメンドデータ１１２に出力する。 The recommendation unit 123 refers to the EA data 111 and specifies the EA identifier related to the relation data 122. The recommendation unit 123 searches the EA data 111 using the name of each entity specified from the relation data 122 as a search key, and identifies an EA identifier similar to the search key. The recommendation unit 123 outputs the specified EA identifier or the data of each item specified from the EA identifier to the recommendation data 112.
　第１の変形例において出力部２４ａは、他の企業などによる過去の改善策から、今回改善が要求されるリレーションに関連するエンタープライズアーキテクチャモデルを、改善策の案として出力する。出力された改善策のうち、今回と同様の課題を解決した改善策が、参照されうる。これにより出力部２４ａは、評価と関連の高いリレーションを出力するとともに、そのリレーションについての解決策を、早期に見いだすことが可能になる。 In the first modification, the output unit 24a outputs the enterprise architecture model related to the relation for which improvement is required this time as a proposal for improvement measures from the past improvement measures by other companies and the like. Among the output improvement measures, the improvement measures that solve the same problems as this time can be referred to. As a result, the output unit 24a can output a relation highly related to the evaluation and can find a solution for the relation at an early stage.
　（第２の変形例）
　図１２を参照して、第２の変形例に係る出力部２４ｂを説明する。出力部２４ｂは、ＥＡデータ（エンタープライズアーキテクチャデータ）１１１のＡｓ－Ｉｓモデルのデータをエンコーダの入力データに設定し、エンタープライズアーキテクチャデータ１１１のＴｏ－Ｂｅモデルのデータをデコーダの入力および正解データとして、予測モデルデータ１１３を学習し、予測モデルデータ１１３を参照して、評価との類似度が高いリレーションのデータを入力データに設定して、Ｔｏ－Ｂｅモデルのデータを出力する。Ｔｏ－Ｂｅモデルのデータは、評価と類似度の高いリレーションの将来と考えられ、改善策となりうる。 (Second modification)
The output unit 24b according to the second modification will be described with reference to FIG. 12. The output unit 24b sets the As-Is model data of the EA data (enterprise architecture data) 111 as the input data of the encoder, and predicts the data of the To-Be model of the enterprise architecture data 111 as the input of the decoder and the correct answer data. The model data 113 is learned, the prediction model data 113 is referred to, the relation data having a high degree of similarity to the evaluation is set as the input data, and the To-Be model data is output. The data of the To-Be model is considered to be the future of relations with high evaluation and similarity, and can be an improvement measure.
　出力部２４ｂは、ＥＡデータ１１１、レコメンドデータ１１２、予測モデルデータ１１３、特定部１２１、レコメンド部１２３ａおよび学習部１２４を備える。 The output unit 24b includes an EA data 111, a recommendation data 112, a prediction model data 113, a specific unit 121, a recommendation unit 123a, and a learning unit 124.
　ＥＡデータ１１１、レコメンドデータ１１２、特定部１２１およびリレーションデータ１２２は、第１の変形例でした通りである。 The EA data 111, the recommendation data 112, the specific unit 121, and the relation data 122 are as in the first modification.
　学習部１２４は、ＥＡデータ１１１を入力として学習し、予測モデルデータ１１３を出力する。学習部１２４は、すでに業務改善が図られた実績のあるＥＡデータ１１１を用いて、教師強制を用いて、Ｓｅｑ２Ｓｅｑ（sequence to sequence）モデルを構築する。図１３に示すように、ＥＡデータ１１１のＡｓ－Ｉｓモデルのデータを、エンコーダ部の入力として設定する。ＥＡデータ１１１のＴｏ－Ｂｅモデルのデータを、デコーダ部の入力および正解データとして設定する。ＥＡデータ１１１のＡｓ－ＩｓモデルのデータおよびＴｏ－Ｂｅモデルのデータは、例えば、ＸＭＬ（Extensible Markup Language）形式で記載される。 The learning unit 124 learns by inputting the EA data 111 and outputs the prediction model data 113. The learning unit 124 constructs a Seq2Seq (sequence to sequence) model by using teacher coercion using the EA data 111 that has already been improved in business. As shown in FIG. 13, the data of the As-Is model of the EA data 111 is set as the input of the encoder unit. The data of the To-Be model of the EA data 111 is set as the input of the decoder unit and the correct answer data. The data of the As-Is model and the data of the To-Be model of the EA data 111 are described in, for example, XML (Extensible Markup Language) format.
　学習部１２４は、最適化アルゴリズムなどをチューニングし、誤差が十分に収束したモデルを用いて予測モデルデータ１１３を生成する。学習部１２４は、ＥＡデータのＡｓ－ＩｓモデルからＴｏ－Ｂｅモデルを予測する予測モデルを学習する。学習部１２４が、例えばＰｙｔｈｏｎのＫｅｒａｓフレームワークを用いて学習する場合、学習の結果得られたｌｏｓｓ値が、誤差になる。ｌｏｓｓ値が限りなく０に近い値となっている場合、誤差が充分に収束したと判定される。 The learning unit 124 tunes the optimization algorithm and the like, and generates the prediction model data 113 using a model in which the error is sufficiently converged. The learning unit 124 learns a prediction model that predicts the To-Be model from the As-Is model of the EA data. When the learning unit 124 learns using, for example, Python's Keras framework, the loss value obtained as a result of the learning becomes an error. When the loss value is as close to 0 as possible, it is determined that the error has sufficiently converged.
　学習部１２４が予測モデルデータ１１３を生成すると、レコメンド部１２３ａは、予測モデルデータ１１３に、Ａｓ－Ｉｓモデルとして、リレーションデータ１２２のリレーション、具体的には評価との類似度が高いリレーションのエンティティ名を入力する。レコメンド部１２３ａは、特定部１２１によって特定されたリレーションにおける改善策として、Ｔｏ－Ｂｅモデルを取得して、出力する。 When the learning unit 124 generates the prediction model data 113, the recommendation unit 123a transfers the prediction model data 113 to the relation of the relation data 122 as an As-Is model, specifically, the entity name of the relation having a high degree of similarity to the evaluation. Enter. The recommendation unit 123a acquires and outputs a To-Be model as an improvement measure in the relation specified by the specific unit 121.
　第２の変形例において出力部２４ｂは、他の企業などによる過去の改善策を学習した予測モデルデータ１１３を参照して、今回改善が要求されるリレーションに関連するＴｏ－Ｂｅモデルを、改善策の案として出力する。これにより出力部２４ｂは、評価と関連の高いリレーションを出力するとともに、そのリレーションについての解決策を、早期に見いだすことが可能になる。 In the second modification, the output unit 24b refers to the prediction model data 113 that has learned the past improvement measures by other companies and the like, and improves the To-Be model related to the relation that is required to be improved this time. Output as a plan. As a result, the output unit 24b can output a relation highly related to the evaluation and can find a solution for the relation at an early stage.
　上記説明した本実施形態の処理装置１は、例えば、CPU（Central Processing Unit、プロセッサ）９０１と、メモリ９０２と、ストレージ９０３（HDD：Hard Disk Drive、SSD：Solid State Drive）と、通信装置９０４と、入力装置９０５と、出力装置９０６とを備える汎用的なコンピュータシステムが用いられる。このコンピュータシステムにおいて、CPU９０１がメモリ９０２上にロードされた処理プログラムを実行することにより、処理装置１の各機能が実現される。なお、CPU９０１と併用して、GPUが用いられても良い。 The processing device 1 of the present embodiment described above includes, for example, a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), and a communication device 904. , A general purpose computer system including an input device 905 and an output device 906 is used. In this computer system, each function of the processing device 1 is realized by executing the processing program loaded on the memory 902 by the CPU 901. A GPU may be used in combination with the CPU 901.
　なお、処理装置１は、１つのコンピュータで実装されてもよく、あるいは複数のコンピュータで実装されても良い。また処理装置１は、コンピュータに実装される仮想マシンであっても良い。 The processing device 1 may be mounted on one computer or may be mounted on a plurality of computers. Further, the processing device 1 may be a virtual machine mounted on a computer.
　処理装置１の処理プログラムは、HDD、SSD、USB（Universal Serial Bus）メモリ、CD (Compact Disc)、DVD (Digital Versatile Disc)などのコンピュータ読取り可能な記録媒体に記憶することも、ネットワークを介して配信することもできる。 The processing program of the processing device 1 can be stored in a computer-readable recording medium such as an HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), DVD (Digital Versatile Disc), or via a network. It can also be delivered.
　なお、本発明は上記実施形態に限定されるものではなく、その要旨の範囲内で数々の変形が可能である。 The present invention is not limited to the above embodiment, and many modifications can be made within the scope of the gist thereof.
　１　処理装置
　２　評価提供装置
　３　評価対象組織
　１１　評価データ
　１２　モデルデータ
　１３　類似度データ
　２１　取得部
　２２　定義部
　２３　類似度算出部
　２４　出力部
　１１１　ＥＡデータ
　１１２　レコメンドデータ
　１１３　予測モデルデータ
　１２１　特定部
　１２２　リレーションデータ
　１２３　レコメンド部
　１２４　学習部
　９０１　CPU
　９０２　メモリ
　９０３　ストレージ
　９０４　通信装置
　９０５　入力装置
　９０６　出力装置 1 Processing device 2 Evaluation provider 3 Evaluation target organization 11 Evaluation data 12 Model data 13 Similarity data 21 Acquisition section 22 Definition section 23 Similarity calculation section 24 Output section 111 EA data 112 Recommendation data 113 Prediction model data 121 Specific section 122 Relations Data 123 Recommendation part 124 Learning part 901 CPU
902 Memory 903 Storage 904 Communication device 905 Input device 906 Output device

Claims (8)

  1. 　評価対象に対する評価を取得する取得部と、
    　評価対象におけるエンティティの各リレーションと、前記評価との類似度を算出する類似度算出部と、
    　各リレーションのうち、前記評価との類似度が高いリレーションを出力する出力部
    　を備える処理装置。     The acquisition department that acquires the evaluation for the evaluation target, and
    A similarity calculation unit that calculates the similarity between each relation of the entity in the evaluation target and the evaluation.
    A processing device including an output unit that outputs a relation having a high degree of similarity to the evaluation among each relation.
  2. 　前記リレーションと前記評価との類似度は、前記リレーションのエンティティ名の合成ベクトルと、前記評価に含まれる単語の合成ベクトルとの、類似度である
    　請求項１に記載の処理装置。     The processing device according to claim 1, wherein the similarity between the relation and the evaluation is the similarity between the synthetic vector of the entity name of the relation and the synthetic vector of the words included in the evaluation.
  3. 　前記取得部は、複数の評価を取得し、
    　前記類似度算出部は、前記複数の評価のそれぞれについて、評価と各リレーションとの類似度を算出し、
    　前記出力部は、前記リレーション毎に、前記類似度が所定値以上の評価の数を出力する
    　請求項１または２に記載の処理装置。     The acquisition unit acquires a plurality of evaluations and obtains a plurality of evaluations.
    The similarity calculation unit calculates the similarity between the evaluation and each relation for each of the plurality of evaluations.
    The processing device according to claim 1 or 2, wherein the output unit outputs the number of evaluations having a similarity of a predetermined value or more for each relation.
  4. 　前記取得部は、複数の評価を取得し、
    　前記類似度算出部は、前記複数の評価のそれぞれについて、評価と各リレーションとの類似度を算出し、
    　前記出力部は、前記リレーション毎に、前記類似度が所定値以上の評価の感情スコアに対応する指標を出力する
    　請求項１または２に記載の処理装置。     The acquisition unit acquires a plurality of evaluations and obtains a plurality of evaluations.
    The similarity calculation unit calculates the similarity between the evaluation and each relation for each of the plurality of evaluations.
    The processing device according to claim 1 or 2, wherein the output unit outputs an index corresponding to an emotion score of evaluation having a similarity of a predetermined value or more for each relation.
  5. 　前記出力部は、エンタープライズアーキテクチャデータから、前記評価との類似度が高いリレーションに関連するデータを取得して出力する
    　請求項１ないし４のいずれか１項に記載の処理装置。     The processing device according to any one of claims 1 to 4, wherein the output unit acquires and outputs data related to a relationship having a high degree of similarity to the evaluation from the enterprise architecture data.
  6. 　前記出力部は、エンタープライズアーキテクチャデータのＡｓ－Ｉｓモデルのデータをエンコーダの入力データに設定し、前記エンタープライズアーキテクチャデータのＴｏ－Ｂｅモデルのデータをデコーダの入力および正解データとして、予測モデルデータを学習し、前記予測モデルデータを参照して、前記評価との類似度が高いリレーションのデータを前記入力データに設定して、Ｔｏ－Ｂｅモデルのデータを出力する
    　請求項１ないし５のいずれか１項に記載の処理装置。     The output unit sets the As-Is model data of the enterprise architecture data as the input data of the encoder, and learns the prediction model data by using the To-Be model data of the enterprise architecture data as the input of the decoder and the correct answer data. In any one of claims 1 to 5, the data of the relation having a high degree of similarity to the evaluation is set as the input data with reference to the prediction model data, and the data of the To-Be model is output. The processing device described.
  7. 　コンピュータが、評価対象に対する評価を取得するステップと、
    　前記コンピュータが、評価対象におけるエンティティの各リレーションと、前記評価との類似度を算出するステップと、
    　前記コンピュータが、各リレーションのうち、前記評価との類似度が高いリレーションを出力するステップ
    　を備える処理方法。     The steps by which the computer obtains an evaluation for the evaluation target,
    A step in which the computer calculates the similarity between each relation of the entity in the evaluation target and the evaluation.
    A processing method in which the computer includes a step of outputting a relation having a high degree of similarity to the evaluation among the relations.
  8. 　コンピュータを、請求項１ないし請求項６のいずれか１項に記載の処理装置として機能させるための処理プログラム。 A processing program for making a computer function as the processing apparatus according to any one of claims 1 to 6.
PCT/JP2020/032016 2020-05-27 2020-08-25 Processing device, processing method and processing program WO2021240832A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2022527476A JP7477791B2 (en) 2020-05-27 2020-08-25 Processing device, processing method, and processing program

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JPPCT/JP2020/020928 2020-05-27
PCT/JP2020/020928 WO2021240686A1 (en) 2020-05-27 2020-05-27 Processing device, processing method, and processing program

Publications (1)

Publication Number Publication Date
WO2021240832A1 true WO2021240832A1 (en) 2021-12-02

Family

ID=78723105

Family Applications (2)

Application Number Title Priority Date Filing Date
PCT/JP2020/020928 WO2021240686A1 (en) 2020-05-27 2020-05-27 Processing device, processing method, and processing program
PCT/JP2020/032016 WO2021240832A1 (en) 2020-05-27 2020-08-25 Processing device, processing method and processing program

Family Applications Before (1)

Application Number Title Priority Date Filing Date
PCT/JP2020/020928 WO2021240686A1 (en) 2020-05-27 2020-05-27 Processing device, processing method, and processing program

Country Status (2)

Country Link
JP (1) JP7477791B2 (en)
WO (2) WO2021240686A1 (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001256251A (en) * 2000-03-08 2001-09-21 Nec Software Chugoku Ltd Device and system for automatically evaluating document information
JP2004046588A (en) * 2002-07-12 2004-02-12 Katsuhiko Inoue Complaint information processing system
JP2007025823A (en) * 2005-07-12 2007-02-01 Fujitsu Ltd Simulation program and simulation method
JP2008287328A (en) * 2007-05-15 2008-11-27 Ntt Data Corp Evaluation device, method, and computer program
JP2011233164A (en) * 2011-07-21 2011-11-17 Mitsubishi Electric Corp Sentence associating system and sentence associating program
WO2016132558A1 (en) * 2015-02-20 2016-08-25 株式会社Ｕｂｉｃ Information processing device and method, and program

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7168334B2 (en) 2018-03-20 2022-11-09 ヤフー株式会社 Information processing device, information processing method and program

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001256251A (en) * 2000-03-08 2001-09-21 Nec Software Chugoku Ltd Device and system for automatically evaluating document information
JP2004046588A (en) * 2002-07-12 2004-02-12 Katsuhiko Inoue Complaint information processing system
JP2007025823A (en) * 2005-07-12 2007-02-01 Fujitsu Ltd Simulation program and simulation method
JP2008287328A (en) * 2007-05-15 2008-11-27 Ntt Data Corp Evaluation device, method, and computer program
JP2011233164A (en) * 2011-07-21 2011-11-17 Mitsubishi Electric Corp Sentence associating system and sentence associating program
WO2016132558A1 (en) * 2015-02-20 2016-08-25 株式会社Ｕｂｉｃ Information processing device and method, and program

Also Published As

Publication number Publication date
JPWO2021240832A1 (en) 2021-12-02
JP7477791B2 (en) 2024-05-02
WO2021240686A1 (en) 2021-12-02

Similar Documents

Publication Publication Date Title
US10430469B2 (en) Enhanced document input parsing
CN103443787B (en) For identifying the system of text relation
Kagdi et al. Assigning change requests to software developers
US9536003B2 (en) Method and system for hybrid information query
AU2018264012B1 (en) Identification of domain information for use in machine learning models
JP4464975B2 (en) Computer apparatus, computer program, and method for calculating the importance of an electronic document on a computer network based on a critique of the electronic document by another electronic document related to the electronic document
US20090077531A1 (en) Systems and Methods to Generate a Software Framework Based on Semantic Modeling and Business Rules
US20150088593A1 (en) System and method for categorization of social media conversation for response management
Wylot et al. Tripleprov: Efficient processing of lineage queries in a native rdf store
US20190026436A1 (en) Automated system and method for improving healthcare communication
US8661004B2 (en) Representing incomplete and uncertain information in graph data
Kagdi et al. Who can help me with this change request?
Pita et al. A Spark-based Workflow for Probabilistic Record Linkage of Healthcare Data.
WO2019016647A1 (en) Automated system and method for improving healthcare communication
Welten et al. DAMS: A distributed analytics metadata schema
US8862609B2 (en) Expanding high level queries
Arch-Int et al. Graph‐Based Semantic Web Service Composition for Healthcare Data Integration
WO2021240832A1 (en) Processing device, processing method and processing program
US20230081891A1 (en) System and method of managing knowledge for knowledge graphs
Iacob et al. MARAM: tool support for mobile app review management.
Eken et al. Predicting defects with latent and semantic features from commit logs in an industrial setting
KR20140034350A (en) Method of personalized detailed clinical model for clinical concept
Kock-Schoppenhauer et al. Practical extension of provenance to healthcare data based on the W3C PROV standard
Pérez Pupo et al. Linguistic data summarization with multilingual approach
Ahmed et al. Ontological Based Approach of Integrating Big Data: Issues and Prospects

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20937884

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022527476

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20937884

Country of ref document: EP

Kind code of ref document: A1