Prediction system, prediction method, and program

Abstract

This prediction system includes: a collection unit that collects a plurality of document data; a classification unit that classifies the document data into numerical data and text data; an input unit that receives input of a keyword serving as an index; an extraction unit that extracts actual measurement data related to the keyword from the numerical data and extracts a description related to future prediction associated with the keyword from the text data; an analysis unit that establishes an approximate expression for the actual measurement data on the basis of the actual measurement data, and sets an index and / or a constant of the approximate expression on the basis of a description related to the future prediction, and generates a prediction formula related to the index; and an output unit that outputs prediction data related to the index on the basis of the prediction formula.

Description

Prediction system, prediction method, and program

[0001] The present invention relates to a prediction system, a prediction method, and a program.

[0002] There is a system that predicts the predicted values ​​of certain indicators. It also predicts how CO2 (carbon dioxide) emissions will change in each country or region in the future based on two patterns: "eventual trends" and "technological progress" (see, for example, Non-Patent Document 1).

[0003] "IEEJ Outlook 2024", Institute of Energy Economics, Japan [Retrieved July 8, 2024], Internet,<https: / / eneken.ieej.or.jp / data / 11380.pdf>

[0004] For example, in the process of planning and formulating a business strategy, when a company needs to quantitatively predict and evaluate the results of the strategy's execution, a simulation using a prediction model for the evaluation index is generally performed. However, when a prediction model or a predicted value does not exist, the current situation is that users collect information and, taking that information into consideration, set up a prediction model or a predicted value themselves.

[0005] Therefore, it has been difficult for users to predict forecast data for a given index based on qualitative descriptions of future predictions.

[0006] The embodiments of the present invention have been made in consideration of the above-mentioned problems, and make it easy for a user to predict forecast data for a predetermined indicator based on a qualitative description of future predictions.

[0007] In order to solve the above problems, a prediction system according to an embodiment of the present invention includes a collection unit that collects multiple pieces of literature data; a classification unit that classifies the literature data into numerical data and text data; an input unit that accepts input of keywords that serve as indicators; an extraction unit that extracts actual measurement data related to the keywords from the numerical data and extracts descriptions relating to future predictions related to the keywords from the text data; an analysis unit that calculates an approximation formula for the actual measurement data based on the actual measurement data and sets an exponent and / or a constant for the approximation formula based on the descriptions relating to the future predictions to generate a prediction formula for the indicator; and an output unit that outputs prediction data for the indicator based on the prediction formula.

[0008] According to an embodiment of the present invention, it becomes easy for a user to predict forecast data for a predetermined indicator based on a qualitative description of future predictions.

[0009] FIG. 1 is a diagram illustrating an example of the configuration of a prediction system according to the present embodiment; FIG. 2 is a diagram illustrating an overview of processing according to the present embodiment; FIG. 3 is a diagram illustrating an overview of generating a prediction formula according to the present embodiment; FIG. 4 is a diagram illustrating an overview of generating a prediction formula according to the present embodiment; FIG. 5 is a flowchart illustrating an example of prediction processing according to the present embodiment; FIG. 6 is a diagram illustrating an example of analysis processing according to Example 1; FIG. 7 is a diagram illustrating an image of a prediction graph according to Example 1; FIG. 8 is a diagram illustrating an example of analysis processing according to Example 2; FIG. 9 is a diagram illustrating an example of a hardware configuration of a computer;

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.

[0011] <Background and Issues> In the process of planning and formulating a business strategy, when a company needs to quantitatively predict and evaluate the results of the strategy's execution, it generally performs simulations using a prediction model for the evaluation indicators. However, when a prediction model or a predicted value does not exist, the current situation is that users collect information and, taking that information into consideration, set up a prediction model or a predicted value themselves.

[0012] For example, Non-Patent Document 1 predicts how CO2 emissions will change in each country or region in the future based on two patterns: "eventual trends" and "technological progress." Conventionally, when predicted values ​​are presented in such documents, users have set future predicted values ​​based on these predicted values.

[0013] However, when only a qualitative description is available regarding future predictions for data of a certain index, it is difficult for a user to set a future predicted value based on the qualitative description regarding future predictions.

[0014] Therefore, in this embodiment, in order to make it easier for a user to predict prediction data regarding a predetermined index based on a qualitative description of future predictions, for example, a configuration such as that shown in FIG. 1 is provided.

[0015] <Configuration Example> Fig. 1 is a diagram showing a configuration example of a prediction system according to this embodiment. In the example of Fig. 1, the prediction system 1 is configured by a prediction device 100. However, this is just an example, and the prediction system 1 may be configured by multiple information processing devices.

[0016] The prediction device 100 is an information processing device having a computer configuration, or a system configured with multiple computers. The prediction device 100 executes a predetermined program to realize, for example, each functional configuration as shown in FIG. 1. In the example of FIG. 1, the prediction system 1 and the prediction device 100 have each functional configuration, such as a collection unit 101, a classification unit 102, an input unit 103, an extraction unit 104, an analysis unit 105, an output unit 106, and an information DB (Database) 110. Note that at least a portion of each of the above functional configurations may be realized by hardware.

[0017] The collection unit 101 executes a collection process for collecting a plurality of literature data. For example, the collection unit 101 collects various literature data with guaranteed reliability issued by the government, public institutions, think tanks, etc. from the Internet, etc.

[0018] The classification unit 102 executes a classification process to classify the document data collected by the collection unit 101 into numerical data 111 and text data 112 and store the data in the information DB 110 .

[0019] The input unit 103 executes an input process for accepting input of keywords serving as indicators. As a specific example, a user specifies keywords such as "CO2 emissions" or "global average temperature." Note that the input unit 103 may be called, for example, a reception unit that accepts input of keywords.

[0020] The extraction unit 104 extracts actual measurement data, etc. related to the keyword from the numerical data 111 classified by the classification unit 102, and performs extraction processing to extract descriptions regarding future predictions related to the keyword from the text data classified by the classification unit 102.

[0021] The analysis unit 105 executes an analysis process (or generation process) to find an approximation formula for the actual measurement data based on the extracted actual measurement data, and to set exponents and constants of the approximation formula based on the extracted description of future predictions, thereby generating a prediction formula for the index. Note that specific examples of the process performed by the analysis unit 105 will be described later.

[0022] The output unit 106 executes an output process to output predicted data related to the index based on the prediction formula generated by the analysis unit 105. For example, the output unit 106 outputs time-series predicted data based on the prediction formula.

[0023] The information DB 110 is a database that stores, for example, the numerical data 111 classified by the classification unit 102 and the text data 112. Note that the information DB 110 may be provided outside the prediction device 100 or outside the prediction system 1 (for example, on the cloud).

[0024] 2 is a diagram for explaining an outline of the process according to this embodiment. This diagram shows an outline of an example of the process executed by the prediction system 1.

[0025] The collection unit 101 collects a variety of documents whose reliability is guaranteed from the Internet or the like (step S1). The collected documents are classified by the classification unit 102 into numerical data 111 and text data 112. Note that documents whose reliability is guaranteed include, but are not limited to, documents issued by the government or public institutions, for example.

[0026] The input unit 103 receives, for example, an input of a keyword serving as an index by the user (step S2).

[0027] Based on the input keyword, the extraction unit 104 extracts actual measurement data related to the keyword from the numerical data 111, and extracts descriptions related to future predictions related to the keyword from the text data 112 (step S3).

[0028] The analysis unit 105 performs regression analysis on the actual measurement data extracted by the extraction unit 104 to find an approximation formula for the actual measurement data. Furthermore, the analysis unit 105 sets an exponent and a constant for the approximation formula based on the description regarding future prediction extracted by the extraction unit 104, and generates a prediction formula for the index.

[0029] 3 and 4 are diagrams for explaining an overview of generation of a prediction formula according to this embodiment. FIG. 3 shows an image of a graph 300 representing a prediction formula for an index, with the horizontal axis representing time and the vertical axis representing the index. The analysis unit 105 performs regression analysis on the actual measurement data extracted by the extraction unit 104. As a result, it is assumed that the approximate formula for the actual measurement data is, for example, y = ax + b (a and b are constants). In this case, the analysis unit 105 can output a graph 301 showing the relationship between time up to the present and the index, for example, as shown in FIG. 3, based on this approximate formula.

[0030] Furthermore, the analysis unit 105 determines the exponents and constants of the approximation formula based on the qualitative description of the future prediction (qualitative description) extracted by the extraction unit 104. For example, the analysis unit 105 determines prediction formulas (examples) of patterns I to V corresponding to the qualitative description using correspondence information 400 as shown in Fig. 4. Furthermore, based on the determined quantitative prediction formula, the analysis unit 105 can output a graph 302 showing the relationship between future time and the index, as shown in Fig. 3, for example.

[0031] As a result, the prediction system 1 according to this embodiment can make it easy for a user to predict prediction data regarding a predetermined index based on a qualitative description regarding future predictions.

[0032] <Processing Flow> Next, a specific processing flow of the information analysis method according to this embodiment will be described.

[0033] 5 is a flowchart showing an example of a prediction process according to this embodiment, which is executed by the prediction system 1 described with reference to FIG.

[0034] In step S501, the collection unit 101 collects a plurality of documents with guaranteed reliability from the Internet, etc. For example, the collection unit 101 collects chart data on the past, present, and future related to a predetermined indicator, qualitative text data related to future predictions, etc.

[0035] In step S502, the classification unit 102 classifies the document data collected by the collection unit 101 into numerical data 111 and text data 112. For example, the classification unit 102 classifies the collected data into numerical data 111 such as graphs or charts, and qualitative text data 112, and stores the data in the information DB 110.

[0036] In step S503, the input unit 103 accepts input of a keyword that serves as an index. For example, the user can input an index that the user wants to extract as an input parameter.

[0037] In step S504, the extraction unit 104 extracts actual measurement data related to the keyword from the numerical data 111, and extracts from the text data 112 descriptions (qualitative descriptions) related to future predictions related to the keyword.

[0038] In step S505, the analysis unit 105 performs regression analysis on the extracted measured data to obtain an approximate expression for the measured data.

[0039] In step S506, the analysis unit 105 uses the obtained approximation formula to set an index and / or a coefficient according to the content of the description regarding future prediction, and obtains a prediction formula regarding the indicator.

[0040] At this time, the user can fine-tune the exponent and / or the constant value while referring to the predicted graph output on the display unit, for example.

[0041] In step S507, the output unit 106 outputs predicted data regarding the index using the prediction formula obtained by the analysis unit 105. For example, the output unit 106 uses the prediction formula to generate and output a predicted value according to a prediction time interval specified by the user.

[0042] <Example> Next, a specific example will be described.

[0043] [Example 1] In Example 1, as a specific example, it is assumed that a user inputs "CO2 emissions" as a keyword serving as an index. For example, in step S503 of Fig. 5, the input unit 103 accepts input of the keyword "CO2 emissions" serving as an index.

[0044] In step S504 of FIG. 5, the extraction unit 104 extracts actual measurement data related to "CO2 emissions" from the numerical data 111 of the information DB 110, and extracts descriptions regarding future predictions related to "CO2 emissions" from the text data 112.

[0045] 5, the analysis unit 105 performs a regression analysis on the data from the past to the present to obtain an approximate equation for the measured data. In the first embodiment, it is assumed that the approximate equation for the measured data of CO2 emissions from the past to the present is y = ax + b (a and b are constants) obtained by the regression analysis.

[0046] In step S506 of FIG. 5, the analysis unit 105 formulates the formula y=a′x in accordance with the description regarding the future prediction extracted by the extraction unit 104. ｔ +b', the exponent t and the constants a' and b' are determined.

[0047] The analysis unit 105 determines a prediction formula from the qualitative description regarding future predictions (qualitative description) extracted by the extraction unit 104, using, for example, the correspondence information 400 shown in Fig. 6. For example, if the qualitative description includes a statement such as "CO2 emissions are expected to increase sharply in the future," the analysis unit 105 selects a prediction formula of pattern I corresponding to the content of the qualitative description from the correspondence information 400. Furthermore, as shown in Fig. 7, for example, the analysis unit 105 calculates a prediction graph 702 from the present onward using the selected prediction formula of pattern I, and combines this with an approximate graph 701 of actual measurement data up to the present to generate a prediction graph 700.

[0048] The user can fine-tune the values ​​of the exponent t and / or the constants a' and b' while viewing the prediction graph 700 displayed on the display. For example, the output unit 106 displays a display screen including the prediction graph 700 as shown in FIG. 7 on a display unit or the like. The input unit 103 also accepts input by the user to change the exponent t and / or the constant a'. Furthermore, the analysis unit 105 recalculates the prediction graph 702 from the present onward in accordance with the change operation accepted by the input unit 103, and updates the prediction graph 700.

[0049] The output unit 106 uses the prediction formula determined in the above process to generate and output prediction data according to, for example, a prediction interval designated by the user.

[0050] In this way, according to the first embodiment, it becomes easy for a user to predict prediction data regarding a predetermined index based on a qualitative description regarding future predictions.

[0051] [Example 2] In Example 2, it is also assumed that the user inputs "CO2 emissions" as a keyword serving as an index. As a result, for example, in step S503 of Fig. 5, the input unit 103 accepts the input of the keyword "CO2 emissions" serving as an index.

[0052] In step S504 of FIG. 5, the extraction unit 104 extracts actual measurement data related to "CO2 emissions" from the numerical data 111 of the information DB 110, and extracts descriptions regarding future predictions related to "CO2 emissions" from the text data 112.

[0053] 5, the analysis unit 105 performs a regression analysis on the data from the past to the present to obtain an approximate equation for the measured data. In the second embodiment, too, it is assumed that the approximate equation for the measured data of CO2 emissions from the past to the present is y = ax + b (a and b are constants) obtained by the regression analysis.

[0054] In step S506 of FIG. 5, the analysis unit 105 formulates the formula y=a′x in accordance with the description regarding the future prediction extracted by the extraction unit 104. ｔ +b', the exponent t and the constants a' and b' are determined.

[0055] The analysis unit 105 determines a prediction formula from the qualitative description of future predictions (qualitative description) extracted by the extraction unit 104, using, for example, the correspondence information 400 shown in FIG. 8. For example, assume that the qualitative description contains a description about CO2 emissions that reads, "After increasing monotonically until 20xx, the change will level off, and it is expected to reach XX billion tons in 2050." In this way, when the time range of the final prediction data is wide and the prediction trend changes midway (when there is an inflection point), the analysis unit 105 selects two patterns, for example, patterns II and III, corresponding to the qualitative description from the correspondence information 400.

[0056] 9 , the analysis unit 105 calculates a prediction graph 902 from the present to an inflection point 904 using the prediction formula of pattern II. The analysis unit 105 also calculates a prediction graph 903 from the inflection point 904, for example, up to the year 2050, using the prediction formula of pattern III. The analysis unit 105 then combines an approximation graph 901 of the actual measurement data up to the present, the prediction graph 902 from the present to the inflection point 904, and the prediction graph 903 from the inflection point 904 to the year 2050 to create a prediction graph 900.

[0057] Furthermore, similar to the first embodiment, the user can adjust the values ​​of the exponent t and / or the constants a', b', c, etc. while viewing the prediction graph displayed on the display.

[0058] The output unit 106 uses the prediction formula determined in the above process to generate and output prediction data according to, for example, a prediction interval designated by the user.

[0059] In this way, according to the second embodiment, it is possible for a user to easily predict forecast data for a predetermined index based on a qualitative description of future predictions. Furthermore, according to the second embodiment, it is also possible to deal with cases where the forecast trend changes midway (when there is an inflection point).

[0060] <Hardware Configuration> The prediction device 100 according to this embodiment has, for example, the hardware configuration of a computer 1000 as shown in Fig. 10. Note that the prediction system 1 may be realized by one or more computers 1000. Furthermore, the computer is not limited to a physical machine and may be, for example, a virtual machine on a cloud.

[0061] Fig. 10 is a diagram showing an example of the hardware configuration of a computer. In the example of Fig. 10, a computer 1000 includes a drive device 1001, an auxiliary storage device 1002, a memory device 1003, a CPU 1004, an interface device 1005, a display device 1006, an input device 1007, and an output device 1008, all of which are interconnected by a bus B. The computer 1000 may further include another processor such as a GPU (Graphics Processing Unit).

[0062] A program for implementing processing on the computer 1000 is provided by a recording medium 1011, such as a CD-ROM or a memory card. When the recording medium 1011 storing the program is set in the drive device 1001, the program is installed from the recording medium 1011 to the auxiliary storage device 1002 via the drive device 1001. However, the program does not necessarily have to be installed from the recording medium 1011, but may be downloaded from another computer via a network. The auxiliary storage device 1002 stores the installed program as well as necessary files, data, etc.

[0063] The memory device 1003 reads and stores the program from the auxiliary storage device 1002 when an instruction to start the program is received. The CPU 1004 realizes, for example, functions related to the prediction device 100 in accordance with the program stored in the memory device 1003. The interface device 1005 is used as an interface for connecting to a communication network or the like. The display device 1006 displays a GUI (Graphical User Interface) or the like according to the program. The input device 1007 is composed of a keyboard, a mouse, buttons, and / or a touch panel, and is used to input various operation instructions. The output device 1008 outputs the results of calculations.

[0064] The CPU 1004 may be another processor such as a DSP (Digital Signal Processor), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array).

[0065] <Effects of the Embodiment> According to the present embodiment, it becomes easy for a user to predict prediction data regarding a predetermined index based on a qualitative description regarding a future prediction.

[0066] For example, a user can easily obtain a prediction formula for obtaining a quantitative predicted value for the future from a qualitative description of past actual values ​​and future predictions.

[0067] Summary of Embodiments This specification discloses at least the following prediction systems, prediction methods, programs, and prediction devices: (1) A prediction system comprising: a collection unit that collects a plurality of literature data; a classification unit that classifies the literature data into numerical data and text data; an input unit that accepts input of keywords serving as indicators; an extraction unit that extracts actual measurement data related to the keywords from the numerical data and extracts descriptions related to future predictions related to the keywords from the text data; an analysis unit that calculates an approximation formula for the actual measurement data based on the actual measurement data and sets an index and / or a constant for the approximation formula based on the descriptions related to future predictions to generate a prediction formula for the indicator; and an output unit that outputs prediction data for the indicator based on the prediction formula. (2) The prediction system described in (1), wherein the output unit displays a display screen including a prediction graph for the indicator; the input unit accepts input of a change operation for the index and / or the constant; and the analysis unit changes the prediction graph for the indicator in response to the change operation. (Clause 3) A prediction method in which a computer executes the following steps: a collection process for collecting multiple pieces of literature data, a classification process for classifying the literature data into numerical data and text data, an input process for accepting input of keywords that serve as indicators, an extraction process for extracting actual measurement data related to the keywords from the numerical data and extracting descriptions related to future predictions related to the keywords from the text data, an analysis process for finding an approximation formula for the actual measurement data based on the actual measurement data and determining an exponent and / or a constant for the approximation formula based on the descriptions related to the future predictions to generate a prediction formula for the indicator, and an output process for outputting prediction data for the indicator based on the prediction formula. (Clause 4) A program, or a storage medium storing a program, that causes a computer to execute the prediction method described in clause 3.(Clause 5) A prediction device comprising: a collection unit that collects multiple pieces of literature data; a classification unit that classifies the literature data into numerical data and text data; an input unit that accepts input of keywords that serve as indicators; an extraction unit that extracts actual measurement data related to the keywords from the numerical data and extracts descriptions related to future predictions related to the keywords from the text data; an analysis unit that finds an approximation formula for the actual measurement data based on the actual measurement data and sets an exponent and / or a constant for the approximation formula based on the descriptions related to future predictions to generate a prediction formula for the indicator; and an output unit that outputs prediction data for the indicator based on the prediction formula.

[0068] Although the present embodiment has been described above, the present invention is not limited to such a specific embodiment, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims.

[0069] REFERENCE SIGNS LIST 1 Prediction system 100 Prediction device 101 Collection unit 102 Classification unit 103 Input unit 104 Extraction unit 105 Analysis unit 106 Output unit 111 Numerical data 112 Text data 1000 Computer

Claims

1. A prediction system comprising: a collection unit that collects multiple pieces of literature data; a classification unit that classifies the literature data into numerical data and text data; an input unit that accepts input of keywords that serve as indicators; an extraction unit that extracts actual measurement data related to the keywords from the numerical data and extracts descriptions related to future predictions related to the keywords from the text data; an analysis unit that calculates an approximation formula for the actual measurement data based on the actual measurement data and sets an exponent and / or a constant for the approximation formula based on the descriptions related to the future predictions to generate a prediction formula for the indicator; and an output unit that outputs prediction data for the indicator based on the prediction formula.

2. The prediction system of claim 1, wherein the output unit displays a display screen including a predicted graph of the index, the input unit accepts input of a change operation for the index and / or the constant, and the analysis unit changes the predicted graph of the index in accordance with the change operation.

3. A prediction method in which a computer executes the following steps: a collection process for collecting multiple pieces of literature data; a classification process for classifying the literature data into numerical data and text data; an input process for accepting input of keywords that serve as indicators; an extraction process for extracting actual measurement data related to the keywords from the numerical data and extracting descriptions related to future predictions related to the keywords from the text data; an analysis process for calculating an approximation formula for the actual measurement data based on the actual measurement data and determining an exponent and / or constant of the approximation formula based on the descriptions related to future predictions to generate a prediction formula for the indicator; and an output process for outputting prediction data for the indicator based on the prediction formula.

4. A program for causing a computer to execute the prediction method according to claim 3.

Images