JP2007133805A - Risk prediction management system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a risk prediction management system capable of recognizing the magnitude of a risk before an accident takes place and immediately following even an ever-changing social situation. <P>SOLUTION: A sentence element dissolving part 8 dissolves text sentence information of a proposal, report and coverage into text vocabulary information, risk points stored in a risk evaluation dictionary DB 9 are allocated to the text vocabulary information to make the total risk points of the undissolved text sentence information its total risk points, and a similarity risk item category dividing part 12 divides the text sentence information into a category group of a similar risk items. The risk prediction management system has a total risk points changing part 13 which makes the highest points in the risk points of the text sentence information representative total risk points of the category group in the category group, and changes the representative total points to the total risk points of text sentence information that is newly input and stored when the total risk point of the newly inputted and stored text sentence information are larger than the representative total risk points of the category group. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a risk prediction management system that can recognize the magnitude of risk in advance before an accident occurs.

  Conventionally, safety activities have been carried out at business sites. It is common to estimate the scale and frequency of anticipated accidents based on proposals and reports on dangerous matters that have been noticed in the office, conduct safety assessments, and implement countermeasures.

  From the viewpoint of predicting danger, as described in Japanese Patent Application Laid-Open No. 2004-240886 (Patent Document 1), in order to reduce the risk associated with an earthquake, the behavior of the furniture at the time of the earthquake is predicted and a countermeasure is proposed. As described in Japanese Patent Application Laid-Open No. 2005-025297 (Patent Document 2), various techniques for avoiding risks associated with the operation of a factory have been proposed.

JP 2004-240886 A JP 2005-025297 A

  However, in the above-mentioned conventional example, the estimated evaluation of the scale and frequency of the expected accident based on the proposal and the report is ambiguous, and it is not possible to immediately follow the changing social situation from time to time, leading to a major accident in the future Possible proposals and reports were overlooked and buried in a large number of proposals and reports, and there was a problem that the proposals and reports were not used effectively.

  The present invention solves the above-mentioned problems, and the purpose of the present invention is to recognize the magnitude of risk in advance before an accident occurs, and to immediately follow the social situation that changes from moment to moment. It is intended to provide a possible risk prediction management system.

  The first configuration of the risk prediction management system according to the present invention for achieving the above object is to provide text information input means for inputting text text information of proposals, reports and news reports related to risk items, and input by the text information input means. Proposed, reported and reported textual text information for each element of When, Where (Where), Who (Who), What (What), Why (Why), and How (How) Sentence element decomposing means for decomposing into text vocabulary information, and text vocabulary information for each element decomposed by the sentence element decomposing means in association with the text sentence information of proposal / report / report before decomposing for each element Vocabulary storage means to divide and memorize and text for each element of When, Where, Where, Who, What, Why, and How Danger points with preset vocabulary information The risk evaluation dictionary storage means classified and stored for each item and the text vocabulary information for each element stored in the vocabulary storage means are assigned a risk score stored in the risk evaluation dictionary storage means before The total risk score calculation means for calculating the total risk score of the text sentence information of the proposal / report / report, and the total risk score of the text text information of the proposal / report / report before the calculation calculated by the total risk score calculation means Is stored as the total risk score of the text sentence information of the proposal / report / report, and for each element obtained by decomposing the text sentence information of the proposal / report / report by the sentence element decomposing means. Similar risk category classification means for classifying into similar risk category categories based on the text vocabulary information, and the similar risk category classification means In the category group of similar risk items classified by the above, the largest total risk score among the total risk score of the text sentence information of the proposal / report / report is set as the representative total risk score of the category group and newly inputted and stored. If the total risk score of the text text information of the proposal / report / report is larger than the representative total risk score of the category group, the representative total risk score is newly entered and stored in the text of the proposal / report / report And a comprehensive risk score changing means for changing the total risk score to sentence information.

  The second configuration of the risk prediction management system according to the present invention is characterized in that, in the first configuration, the total risk score is composed of a risk scale score and a risk frequency score.

  Further, the third configuration of the risk prediction management system according to the present invention is that the total risk score of the text text information of the proposal / report / report newly input and stored in the first configuration is already input and stored. If the representative total risk score in the similar risk item category is exceeded, all the total risk points belonging to the similar risk item category already input and stored are newly input and stored in the proposal / report.・ It is characterized by the fact that it is replaced with the total risk score of the textual text information of the report.

  According to the first configuration of the risk prediction management system according to the present invention, when the text element information of the proposal / report / report regarding the dangerous matter input by the text information input means is read by the sentence element decomposition means, Where (Where), Who (Who), What (What), Why (Why), How (How) Decompose the text vocabulary information for each element of each element, and divide the text vocabulary information for each element It is possible to divide into each element in association with the text sentence information of the proposal / report before decomposition and store it in the vocabulary storage means.

  On the other hand, the risk assessment dictionary storage means includes when, where (Where), who (Who), what (What), why (Why), and how (How). Text vocabulary information is classified and stored for each preset risk score.

  Then, the total risk score calculation means assigns the risk score stored in the risk evaluation dictionary storage means to the text vocabulary information for each element of the syntax stored in the vocabulary storage means, and proposes, reports, The total risk score of the text text information of the news report can be calculated and stored in the total risk score storage means as the total risk score of the text text information of the proposal / report / report.

  On the other hand, the text category information of proposal / report / report is classified into similar risk category categories based on the text vocabulary information for each element of the syntax decomposed by the text element decomposition means by the similar risk category classification means. By classifying and changing the total risk score, in the classified similar risk category, the largest total risk score among the total risk scores in the text text information of proposal / report / report is set as the representative total risk score in that category. If the total risk score of the newly entered and stored text text information of the proposal / report / report is greater than the representative total risk score in the similar risk category, the representative total risk score is newly input and stored. It can be changed to the total risk score of the text text information of proposed proposals, reports and reports.

  Further, according to the second configuration of the risk prediction management system according to the present invention, the total risk score is composed of the risk scale score and the risk frequency score, so that the overall risk score is evaluated from the viewpoint of both the risk scale and the risk frequency. It can be a risk score.

  Further, according to the third configuration of the risk prediction management system of the present invention, in the total risk score changing means, the text text information of proposals / reports / reports belonging to the similar risk item category already input and stored By substituting all of the total risk points, the risk level of the entire group of similar risk item categories can be made to stand out.

  As a result, before the occurrence of an accident, especially a serious accident, it is possible to recognize the level of risk of textual text information in proposals, reports, and news reports in advance, and to immediately follow the changing social situation. I can do it.

  An embodiment of the risk prediction management system according to the present invention will be specifically described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the control system of the risk prediction management system according to the present invention, FIG. 2 is a flowchart showing the operation of the risk prediction management system according to the present invention, and FIG. FIG. 4 is a diagram showing a state in which text document information is decomposed into text vocabulary information for each element of the syntax to calculate a risk scale score and a risk frequency score. FIG. FIG. 5 is a diagram showing how the frequency score is updated, FIG. 5 is a diagram showing how the risk score of past cases is recalculated, FIG. 6 is a diagram showing an example of a daily action check screen, and FIG. FIG. 8 is a diagram illustrating an example of a risk prediction proposal screen, FIG. 8 is a diagram illustrating an example of an abnormality report screen, FIG. 9 is a diagram illustrating an example of a risk scale selection screen, and FIG. 10 is an example of an occurrence frequency selection screen. FIG. FIG. 12 is a diagram showing an example of a high-risk case search screen, FIG. 12 is a diagram showing an example of an improvement proposal screen, and FIG. 13 is a diagram showing an example of a change management screen.

  In FIG. 1, reference numeral 1 denotes a personal computer (hereinafter referred to as “personal computer”) that constitutes text information input means for inputting text text information on proposals, reports, and news reports regarding dangerous matters. 2 is an external case server device that provides news case examples related to external dangerous matters including public information provided by public news organizations as textual text information on the Internet 3 as a communication means. Also serves as text information input means for inputting information.

  Proposed / reported textual text information related to dangerous matters input from the personal computer 1 at each in-house factory is sent to the risk prediction management server device 5 via the in-house LAN (Local Area Network) 4, while the external case server device. 2 is sent to the risk prediction management server device 5 via the Internet 3 and the firewall 7.

  The text text information of the proposal / report / report input / acquired by the personal computer 1 and the external case server apparatus 2 is sent to the text element decomposition unit 8 serving as the text element decomposition means provided in the risk prediction management server apparatus 5 at any time. (When), Where (Where), Who (Who), What (What), Why (Why), and How (How) are decomposed into text vocabulary information for each element of the syntax.

  For example, as shown in Fig. 3 (a), "A high-pressure gas tank exploded at factory B. According to the fire department, the main cause is corrosion of cooling water piping. The company has caused the same type of accident five years ago. When text text information such as “” is input, as shown in FIG. 3B, each syntax division is “When”, “Where”, “Who”. , "What," "Why," and "How", the text text of the proposal / report / report before breaking down into text vocabulary information for each element of the syntax The information is stored and stored in a case information management database (hereinafter referred to as “case information management DB”) 6 serving as a vocabulary storage unit in association with information.

  On the other hand, in the risk assessment dictionary database (hereinafter referred to as “danger assessment dictionary DB”) 9 serving as a risk assessment dictionary storage means, when, where (Where), who (Who), what (What), why (Why), how the text vocabulary information for each element of the syntax of (How) is classified and stored for each preset risk score.

  The risk score of the present embodiment is composed of a risk scale score and a risk frequency score. For example, as shown in FIG. 3C, “fire station”, “high pressure gas”, “cooling water”, “corrosion”. , “Explosion” and “accident” are set as “300” points, “50” points, “5” points, “30” points, “800” points, “50” points, respectively. The risk frequency score of “5 years ago” is set to “5”.

  A total risk score calculation unit 10 serving as a total risk score calculation unit provided in the risk prediction management server device 5 is provided for text vocabulary information for each element of the syntax stored in the case information management DB 6 serving as a vocabulary storage unit. A risk score composed of a risk scale score and a risk frequency score stored in the risk evaluation dictionary DB 9 is assigned to calculate a total risk score of the text sentence information of the proposal / report / report before the disassembly.

  In the present embodiment, when the total risk score is calculated, the total risk score score set in advance in the text vocabulary information for each element of the syntax is multiplied by the risk frequency score. That is, the total risk score of the text sentence information shown in FIG. 3 is calculated as {total number of danger scale points “1235 points”} × {risk frequency score “5 points”} = 6175 points.

  In this way, the total risk score of the text sentence information of the proposal / report / report text sentence information calculated by the total risk score calculation unit 10 is the total risk score of the proposal / report / report text sentence information. It is stored and stored in a total risk score database (hereinafter referred to as “total risk score DB”) 11 serving as a score storage means.

  On the other hand, the text text information of the proposal / report / report is based on the text vocabulary information for each element of the syntax decomposed by the text element decomposition section 8, and the similar risk category classification section 12 serving as a similar risk category classification means. Into similar risk categories. In the risk assessment dictionary DB 9, a keyword and a similarity threshold for determining whether or not it is similar to the category group are set in advance for each category of each risk item. For example, the text shown in FIG. The text vocabulary information for each element of the syntax included in the sentence information is checked for matches / mismatches with keywords set in advance for each category of each dangerous matter, and the number of matched text vocabulary information is shown in FIG. It is recorded in the similarity column 22 of a). Then, when the number of similar scores exceeds a preset similarity threshold, the similar risk items are classified into category groups and stored in the case information management DB 6.

  For example, as shown in FIG. 4 (a), “high pressure gas tank liquid leakage”, “high pressure gas tank odor”, and “high temperature gas tank temperature rise” having the highest similarity have the highest similarity. “Liquid leakage in gas tank” is selected as the category group of the text sentence information shown in FIG. In this embodiment, the similarity threshold is set to “5”.

  The total risk score changing unit 13 provided as a means for changing the total risk score provided in the risk prediction management server device 5 is a group of similar risk items classified by the similar risk item category classifying unit 12, and proposes / reports / reports. The total risk score of the text text information of the text is the largest total risk score of the category group as the representative total risk score of the category group. If it is larger than the representative total risk score, the representative total risk score is changed to the total risk score of the text sentence information of the proposal / report / report newly input and stored.

  That is, the risk scale score “150” and the risk frequency score “3” are set as the representative total risk score of the “high-pressure gas tank liquid leak” category group shown in FIG. As shown, the risk scale score “1235” and the risk frequency score “5” are set as the total risk score of the newly entered and stored text text information of the proposal / report / report, and these risk scores are As shown in FIG. 4B, if the representative total risk score (risk scale score “150”, risk frequency score “3”) of the category group of “high-pressure gas tank leakage” is larger than that, The risk scale score and the risk frequency score, which are score points, are changed into a risk scale score “1235” and a risk frequency score “5”, which are the total risk scores of the text text information of proposals, reports, and news reports that are newly input and stored. change.

  At this time, for example, when only one of the risk scale score or the risk frequency score is larger than one of the risk scale score of the representative comprehensive risk score or the risk frequency score as the total risk score of the text text information of proposal / report / report Changes only one of the larger risk scale score or the risk frequency score, and the smaller risk score becomes the total risk score as it is.

  Further, as shown in FIG. 5, when the representative total risk score is changed, the total risk score changing unit 13 provides text text information of proposals / reports / reports belonging to the similar risk category already input and stored. All of the total risk points are replaced with the changed representative total risk score.

  In other words, as shown in FIG. 5 (a), the text recorded in the proposal / report / report text text newly entered and stored in the risk score column of the category "leakage of high-pressure gas tank" The total risk score of {1235 points, which is the total sum of the risk scale scores} × {5 points, which is the risk frequency score} = 6175 points.

  Then, based on the total risk score changed by the total risk score change unit 13, as shown in FIG. 5 (b) on the display screen of the personal computer 1 serving as the output means, the proposal / report / report is arranged in descending order of the total risk score. The text text information is output side by side.

Next, using the risk prediction management system according to the present invention will be described with reference to FIG. First, in step S ₁ in FIG. 2, the operator checks for items to be checked routinely on daily activities check screen 14 shown in FIG. 6 by using the personal computer 1. In the case of performing the risk prediction proposal, click the danger prediction proposal button 14a of the day-to-day behavior check screen 14, it moves to the risk prediction proposal screen 15 shown in FIG. 7, to enter a proposal by the text sentence information (step S ₂ ).

Further, when an abnormality report is made, the abnormality report button 14b on the daily action check screen 14 in FIG. 6 is clicked to move to the abnormality report screen 16 shown in FIG. S _3).

As described above with reference to FIG. 3 (b), the text element information of the proposal contents or abnormal contents sent to the risk prediction management server device 5 via the in-house LAN 4 is sent to the When, where (Where), who (Who), what (What), why (Why), and how (How), the sentence is analyzed into text lexical information for each element of the syntax. (Step S ₄ ), as described above with reference to FIG. 3 (c), for the text vocabulary information for each element stored in the case information management DB 6, which is a vocabulary storage means, by the total risk score calculation unit 10. The scale frequency score calculation is performed to assign the risk score composed of the risk scale score and the risk frequency score stored in the risk evaluation dictionary DB9, and to calculate the total risk score of the text sentence information of the proposal / report / report before the decomposition (step) S ₅ ).

Next, as described above with reference to FIG. 4 (a), the textual sentence information is converted by the similar risk item category classification unit 12 based on the text vocabulary information for each element of the syntax decomposed by the sentence element decomposition unit 8. Then, a category search for similar risk items classified into a category group of similar risk items is performed (step S ₆ ).

Next, as shown in FIG. 4B and described above, the total risk score changing unit 13 determines the total risk score of the text sentence information in the category group of similar risk items classified by the similar risk item category classification unit 12. If the total risk score of the newly entered text sentence information is larger than the representative total risk score of the category group, the representative total risk score of the category group is set as the representative total risk score of the category group. The scale frequency score is updated to change the risk score to the total risk score of the text sentence information newly inputted and stored (step S ₇ ).

Next, as shown in FIG. 5 and described above, the total risk score changing unit 13 includes the total risk of all categories included in the category group that has been changed to the total risk score of the text sentence information that is newly input and stored. The risk score recalculation is performed to change the score to the total risk score of the text sentence information newly input and stored (step S ₈ ).

Meanwhile, the has function of artificially modified is prepared autoconfiguration category similar hazards that performed in step S _6, performs a correction input of the scale categories on dangerous scale selection screen 17 shown in FIG. 9 (step S _9), can be corrected input frequency category on frequency selection screen 18 shown in FIG. 10 (step S _10).

Then, a risk calculation is performed to calculate a total risk score based on the risk scale score of the scale category and the risk frequency score of the frequency category that are artificially selected in steps S ₉ and S ₁₀ (step S ₁₁ ). In this embodiment, it is calculated by {total risk score = danger scale score × danger frequency score}.

Next, a high risk case search is performed on the high risk case search screen 19 shown in FIG. 11 (step S ₁₂ ). The high-risk case search screen 19 displayed on the display screen of the personal computer 1 serving as the output means has been changed by the total risk score changing unit 13 or has been artificially changed in steps S ₉ and S _10. Based on the score, the text text information of the proposal / report / report regarding the dangerous matter is arranged and output in descending order of the total risk score.

Then, an improvement proposal is made using the improvement proposal screen 20 shown in FIG. 12 (step S ₁₃ ), and change management is performed using the change management screen 21 shown in FIG. 13 (step S ₁₄ ).

  As a result, it is possible to recognize in advance the risk level of textual text information for proposals, reports, and reports before an accident occurs, and to immediately follow the changing social situation.

  As an application example of the present invention, the present invention can be applied to a risk prediction management system that can recognize the magnitude of risk in advance before an accident occurs.

It is a block diagram which shows the structure of the control system of the danger prediction management system which concerns on this invention. It is a flowchart which shows operation | movement of the danger prediction management system which concerns on this invention. It is a figure which shows a mode that text document information is decomposed | disassembled into the text vocabulary information for every element of a syntax by a sentence element decomposition | disassembly means, and a risk scale score and a risk frequency score are calculated. It is a figure which shows a mode that the category of a similar dangerous matter is searched and a dangerous scale score and a dangerous frequency score are updated. It is a figure which shows a mode that the risk score of the past example is recalculated. It is a figure which shows an example of a daily activity check screen. It is a figure which shows an example of a danger prediction proposal screen. It is a figure which shows an example of an abnormality report screen. It is a figure which shows an example of a dangerous scale selection screen. It is a figure which shows an example of the occurrence frequency selection screen. It is a figure which shows an example of a high risk example search screen. It is a figure which shows an example of an improvement proposal screen. It is a figure which shows an example of a change management screen.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Personal computer 2 ... External case server apparatus 3 ... Internet 4 ... Internal LAN
5 ... Risk prediction management server device 6 ... Case information management DB
7 ... Firewall 8 ... Text element decomposition unit 9 ... Risk assessment dictionary DB
10 ... Total risk score calculation section
11… Danger point DB
12 ... Similar risk category classification
13… Total risk score change section
14 ... Daily behavior check screen
14a… Danger prediction button
14b… Abnormality report button
15… Danger prediction proposal screen
16… Abnormality report screen
17… Danger scale selection screen
18 ... Frequency selection screen
19… High-risk case search screen
20 ... Improvement proposal screen
21 ... Change management screen
22… Similarity column

Claims (3)

Text information input means for inputting text text information on proposals, reports and news reports on dangerous matters;
Proposal / report / report text text information entered by the text information input means, when, where (Where), who (Who), what (What), why (Why), how Sentence element decomposition means that decomposes into text vocabulary information for each element of (How),
Vocabulary storage means for storing the text vocabulary information for each element decomposed by the sentence element decomposition means in association with the text sentence information of the proposal / report / report before the decomposition and storing for each element;
Text vocabulary information for each element of When, Where (Where), Who (Who), What (What), Why (Why), and How (How) is classified by preset risk score A risk assessment dictionary storage means stored as
For the text vocabulary information for each element stored in the vocabulary storage means, the risk score stored in the risk evaluation dictionary storage means is assigned, and the total risk score of the text sentence information of the proposal / report / report before the decomposition. Comprehensive risk score calculation means for calculating
Total risk score storage means for storing the total risk score of the text sentence information of the proposal / report / report text sentence information calculated by the total risk score calculation means as the total risk score of the proposal / report / report text sentence information When,
Similar risk item category classification means for classifying the text sentence information of the proposal / report / report into similar risk category categories based on the text vocabulary information for each element decomposed by the sentence element decomposition means;
In the category group of similar risk items classified by the similar risk item category classification means, the largest total risk score among the total risk scores of the text text information of the proposal / report / report is the representative total risk score of the category group. When the total risk score of the newly entered and stored text sentence information of the proposal / report / report is larger than the representative total risk score of the category group, the representative total risk score is newly input and stored. Total risk score changing means for changing to the total risk score of text text information of proposals, reports and reports,
A risk prediction management system characterized by comprising: The risk prediction management system according to claim 1, wherein the total risk score includes a risk scale score and a risk frequency score. When the total risk score of the newly entered and stored text text information of the proposal / report / report exceeds the representative total risk score in the similar risk category already input and stored, it is already input and stored. 2. The risk according to claim 1, wherein all the total risk points belonging to the similar risk item category are replaced with the total risk score of the newly input and stored text text information of the proposal / report / report. 3. Predictive management system.