Establishment of recommendation model for change risk keywords and method and system for recommendation

Abstract

The present invention relates to the field of change management technology, and discloses a method for establishing a keyword recommendation model, an establishment system, and a method for recommending keywords for evaluating change risks and an establishment system. The establishment method comprises: based on change instances and change accidents, identifying multiple risk factors related to change operations that lead to preset change consequences; extracting a first keyword group from the multiple risk factors, the first keyword group including multiple first keywords; and establishing the keyword recommendation model based on multiple first specific keywords with preset correlations in the first keyword group. The present invention can comprehensively, objectively and intelligently recommend keywords for evaluating change risks, thereby improving the accuracy and efficiency of change risk assessment, and further realizing intelligent change risk management.

Description

Technical Field

[0001] The present invention relates to the technical field of change management, and in particular to a method and system for establishing a keyword recommendation model, and a method and system for recommending keywords for evaluating change risks. Background Art

[0002] Change management in petrochemical enterprises involves flammable, explosive, toxic and harmful materials and products, and involves multiple disciplines such as equipment, instrumentation, and electrical engineering, as well as complex public engineering systems. As an important part of risk management, it is increasingly receiving national attention. Changes are characterized by relatively hidden dangers, complex construction processes, and long latent periods of impact after changes. Major equipment and facility change accidents caused by inadequate change process management have caused huge casualties, which further exposes the fact that change management is still a weak link in risk management. In response to the problems of undetailed change management systems and inadequate change procedures, some development companies have developed change management modules, thereby realizing process-based control of change management, which has been introduced and used by many domestic companies.

[0003] However, the application of the change management library only stays at the level of electronic change management records, and the related change risk identification and assessment also mainly rely on subjective experience and manual extraction. Therefore, the current change management technology still has problems such as incomplete change risk identification, strong subjectivity in change risk assessment, and low assessment accuracy and efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide a method for establishing a keyword recommendation model, an establishment system, a method for recommending keywords for evaluating change risks, and an establishment system, which can comprehensively, objectively and intelligently recommend keywords for evaluating change risks, thereby improving the accuracy and efficiency of change risk assessment, and further realizing intelligent change risk management.

[0005] In order to achieve the above-mentioned objectives, the first aspect of the present invention provides a method for establishing a keyword recommendation model, the method comprising: based on change instances and change accidents, identifying multiple risk factors related to change operations that lead to preset change consequences; extracting a first keyword group from the multiple risk factors, the first keyword group including multiple first keywords; and establishing the keyword recommendation model based on multiple first specific keywords with preset correlations in the first keyword group.

[0006] Preferably, the identification of multiple risk factors related to the change operation that leads to the preset change consequence includes: inferring the uncontrolled link of the risk evolution of the preset change consequence based on the evolutionary time series relationship and evolution law of the change accident and the change instance; and based on the change instance, identifying the multiple risk factors associated with the inferred uncontrolled link by using a logic tree analysis method.

[0007] Preferably, extracting the first keyword group from the multiple risk factors includes: segmenting the multiple risk factors to obtain multiple words; calculating the importance factor of each of the multiple words in a specific document in a corpus, wherein the corpus is formed by the change instances and the change accidents; and screening out words whose importance factors are greater than or equal to an importance factor threshold from the multiple words, and adding the screened words to the first keyword group.

[0008] Preferably, calculating the importance factor of each of the plurality of words in a specific document in the corpus comprises: using the following weighting algorithm to calculate the importance factor of each of the plurality of words in a specific document in the corpus, W tf (i,j)=tf(i,j)×idf(i), where tf(i,j) is the frequency of word i appearing in a specific document j; is the weight of word i appearing in a specific document j, and N is the total number of documents in the corpus.

[0009] Preferably, after executing the step of extracting the first keyword group from the multiple risk factors, the establishment method also includes: classifying the multiple first keywords according to the characteristics of the multiple first keywords in the first keyword group to obtain the multiple first keywords of different classifications; and constructing a change keyword library with the multiple first keywords of different classifications.

[0010] Preferably, the establishing of the keyword recommendation model comprises: constructing the keyword recommendation model in the form of a graph database based on a plurality of first specific keywords of different categories with preset associations.

[0011] Through the above technical scheme, the present invention creatively identifies multiple risk factors related to change operations that lead to specific change consequences comprehensively and objectively based on change instances and change accidents, then extracts multiple keywords based on the identified risk factors, and then establishes a corresponding keyword recommendation model based on multiple specific keywords with certain correlations among the extracted keywords. As a result, keywords for evaluating change risks can be intelligently recommended based on the established keyword recommendation model, thereby improving the accuracy and efficiency of change risk assessment, and further realizing intelligent change risk management.

[0012] A second aspect of the present invention provides a keyword recommendation method for evaluating change risks, the recommendation method comprising: determining the similarity between a change request and each of a plurality of keyword recommendation models obtained according to the keyword recommendation model establishment method; and when the similarity between the change request and a specific keyword recommendation model among the plurality of keyword recommendation models is greater than or equal to a preset similarity threshold, the specific keyword recommendation model recommends a plurality of first specific keywords corresponding to the change request.

[0013] Preferably, the recommendation method also includes: when the similarity between the change request and any one of the multiple keyword recommendation models is less than the preset similarity threshold, matching the change request with multiple second keywords in the second keyword group in the change keyword library to obtain multiple second specific keywords corresponding to the change request, wherein the multiple second keywords are extracted based on the change instance; and recommending multiple second specific keywords corresponding to the change request.

[0014] Preferably, the recommendation method also includes: determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords based on the conditional dependency relationship between the recommended multiple first specific keywords or the multiple second specific keywords and the change request, and accordingly, recommending the multiple first specific keywords or the multiple second specific keywords corresponding to the change request includes: according to the recommendation priority of the multiple first specific keywords or the multiple second specific keywords, screening and recommending the first specific keywords or the second specific keywords with the recommendation priority ranked in the front preset position.

[0015] Preferably, determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords includes: calculating the conditional probability of each of the outputted multiple first specific keywords or the multiple second specific keywords being recommended when the change request occurs based on the conditional probability formula; and determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords according to the conditional probability of each of the multiple first specific keywords or the multiple second specific keywords being recommended when the change request occurs.

[0016] Preferably, determining the similarity between the change request and each of the multiple keyword recommendation models obtained according to the keyword recommendation model establishment method includes: determining the vector corresponding to the change request; determining the vector corresponding to each of the multiple keyword recommendation models; and determining the similarity between the change request and each of the multiple keyword recommendation models based on the cosine similarity between the vector corresponding to the change request and the vector corresponding to each of the multiple keyword recommendation models.

[0017] Through the above technical scheme, the present invention creatively determines the similarity between the change request and each of the multiple keyword recommendation models obtained according to the above keyword recommendation model establishment method. If the similarity between the change request and the specific keyword recommendation model is greater than or equal to the preset similarity threshold, the specific keyword recommendation model outputs the corresponding multiple keywords. Therefore, the change request can be intelligently risk evaluated based on the output multiple keywords, thereby improving the accuracy and efficiency of the change risk assessment, and then realizing intelligent change risk management.

[0018] The third aspect of the present invention also provides a system for establishing a keyword recommendation model, the system comprising: a risk factor identification device for identifying multiple risk factors related to change operations that lead to preset change consequences based on change instances and change accidents; a keyword extraction device for extracting a first keyword group from the multiple risk factors, the first keyword group including multiple first keywords; and a model establishment device for establishing the keyword recommendation model based on multiple first specific keywords with preset correlations in the first keyword group.

[0019] The specific details and benefits of the system for establishing the keyword recommendation model provided by the present invention can be found in the description of the method for establishing the keyword recommendation model mentioned above, which will not be repeated here.

[0020] The fourth aspect of the present invention also provides a keyword recommendation system for evaluating change risks, the recommendation system comprising: a similarity determination device for determining the similarity between a change request and each of a plurality of keyword recommendation models acquired according to the keyword recommendation model establishment system; and a keyword output device for recommending a plurality of first specific keywords corresponding to the change request by the specific keyword recommendation model when the similarity between the change request and a specific keyword recommendation model among the plurality of keyword recommendation models is greater than or equal to a preset similarity threshold.

[0021] The specific details and benefits of the keyword recommendation system for evaluating change risk provided by the present invention can be found in the description of the keyword recommendation method for evaluating change risk, which will not be repeated here.

[0022] The fifth aspect of the present invention also provides a machine-readable storage medium, on which instructions are stored, and the instructions are used to enable a machine to execute the method for establishing the keyword recommendation model and / or the method for recommending keywords for evaluating change risks.

[0023] Other features and advantages of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention but do not constitute a limitation of the present invention. In the accompanying drawings:

[0025] Figure 1 is a flow chart of a method for establishing a keyword recommendation model provided in Embodiment 1 of the present invention;

[0026] Figure 2 is a flowchart of extracting a first keyword group from the multiple risk factors provided by the second embodiment of the present invention;

[0027] Figure 3 is a schematic diagram of a keyword recommendation model provided in Embodiment 2 of the present invention;

[0028] Figure 4 is a flowchart of a method for recommending keywords for evaluating change risk provided by Embodiment 3 of the present invention;

[0029] Figure 5 is a flowchart of determining the similarity between a change request and a keyword recommendation model provided by Embodiment 6 of the present invention;

[0030] Figure 6 is a flowchart of a keyword recommendation process for evaluating change risk provided by Embodiment 6 of the present invention;

[0031] Figure 7 is a structural diagram of a system for establishing a keyword recommendation model provided in Embodiment 8 of the present invention;

[0032] Figure 8 is a structural diagram of a keyword recommendation system based on a keyword recommendation model provided in Embodiment 8 of the present invention; and

[0033] Fig. 9 This is a structural diagram of a keyword recommendation system for evaluating change risk provided by Embodiment 9 of the present invention.

[0034] Description of Reference Numerals

[0035] 10 Risk factor identification device 20 Keyword extraction device

[0036] 30 Model building device 40 Keyword library building equipment

[0037] 50 Similarity determination device 60 First recommendation device

[0038] 70 Priority determination device 80 Change accident acquisition device

[0039] 90 Change instance acquisition device DETAILED DESCRIPTION

[0040] The following nine embodiments are used to describe the specific implementation of the present invention in detail in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0041] Embodiment 1

[0042] Figure 1 FIG. 1 is a flow chart of a method for establishing a keyword recommendation model provided in Embodiment 1 of the present invention. Figure 1 As shown, the establishment method may include: step S101, based on change instances and change accidents, identifying multiple risk factors related to change operations that lead to preset change consequences; step S102, extracting a first keyword group from the multiple risk factors, the first keyword group including multiple first keywords; and step S103, establishing the keyword recommendation model based on multiple first specific keywords with preset correlation in the first keyword group.

[0043] The change operation that leads to the preset change consequence may be: Figure 3 The process card changes that lead to equipment damage shown in the figure may have corresponding risk factors such as failure to inform the operator, display error, operator unfamiliarity, operator not receiving training, misoperation, etc. The preset correlation may include: causal correlation (such as Figure 3 The process card shown in the figure was changed, resulting in display errors; the operator was not informed, resulting in misoperation; the equipment was damaged due to misoperation, etc.) and the attribution relationship (such as Figure 3 The control measures for misoperation shown include timely notification, regular maintenance and inspection, team member training, etc.).

[0044] Therefore, keywords for assessing change risks can be intelligently recommended based on the established keyword recommendation model, thereby improving the accuracy and efficiency of change risk assessment and further realizing intelligent change risk management.

[0045] Embodiment 2

[0046] Based on the first embodiment, the description contents involved in steps S101 to S103 may be further refined.

[0047] Before executing step S101, the establishment method may further include: obtaining the change instance and the change accident. Specifically, based on domestic accident reports and foreign databases (EMARS, CSB, FKD, ARIA), Figure 8 The change accident acquisition device 80 in the embodiment constructs a change accident case library containing 5 categories and 23 subcategories of information including basic accident information, accident impact, change information, accident video, and accident report. The case library architecture is shown in Table 1. In addition, based on the enterprise change instance and HAZOP analysis report, the change accident case library is adopted. Figure 8 The change instance acquisition device 90 constructs a 6-level change instance library including change name, change type (for example, 3 change types), applicable enterprise, applicable department, device / area and change control table (keywords, risk factors, possible causes, possible consequences, control measures).

[0048] Table 1 Change accident case library architecture

[0049]

[0050] For step S101, the identification of multiple risk factors related to the change operation leading to the preset change consequence may include: inferring the uncontrolled link of the risk evolution of the preset change consequence based on the evolutionary time sequence relationship and evolution law of the change accident and the change instance; and identifying the multiple risk factors associated with the inferred uncontrolled link by logic tree analysis based on the change instance. In other words, the purpose of step S101 is to identify the change risk.

[0051] For example, if change incident A is caused by not informing the operator of the change (e.g. Figure 3 Changes to process cards in the process) may lead to change accidents (for example, Figure 3 ), then according to the scenario of the change accident A and the evolutionary time sequence relationship and evolution law of the change instance A' (for example, the process card change instance), the preset change consequence (for example, Figure 3The uncontrolled links of the risk evolution of equipment damage in the process may include the following three points: failure to inform the operator; failure to train the operator; and equipment damage caused by operator error. Then, based on the change accident A and the corresponding change instance A' (for example, the process card change instance), the multiple risk factors associated with the inferred uncontrolled links are analyzed by logical tree analysis. For example, for the link of not informing the operator, the corresponding risk factor may be the operator's unfamiliarity; for the link of not training the operator, the corresponding risk factor may be error operation; and for the link of equipment damage caused by operator error operation, the link may be equipment overpressure. Among them, the risk factors of operator unfamiliarity, error operation, and equipment overpressure can be directly determined based on the uncontrolled link and combined with the change accident and / or change instance; while the risk factors such as display error, insufficient personnel ability, and inadequate valve switch status cannot be determined based on the uncontrolled link and the change accident. At this time, it is necessary to combine the change accident and / or change instance and the logical relationship between other preset factors to be determined. Therefore, the multiple risk factors determined by the above method are more comprehensive and more reasonable.

[0052] For step S102, Figure 2 As shown, extracting the first keyword group from the multiple risk factors may include steps S201-S203.

[0053] Step S201: segment the multiple risk factors to obtain multiple words.

[0054] The multiple risk factors may be segmented by any reasonable existing segmentation technology (e.g., "Jieba" segmentation technology), that is, a continuous sequence of characters may be recombined into a sequence of words according to certain specifications. For example, after segmenting "pipeline increase valve", the words "pipeline", "increase" and "valve" may be obtained.

[0055] Step S202 , calculating the importance factor of each of the plurality of words in a specific document in the corpus.

[0056] The corpus is formed by the change instances and the change accidents. Specifically, the corpus may include a change instance library and a change accident case library; and the importance factor may be used to measure the importance of word i in document j.

[0057] For step S202, calculating the importance factor of each of the multiple words in a specific document in the corpus may include: using the following weighting algorithm (i.e., term frequency-inverse document frequency index (TF-IDF) weighting method) to calculate the importance factor of each of the multiple words in a specific document in the corpus,

[0058] W tf(i,j)=tf(i,j)×idf(i),

[0059] Where tf(i, j) is the frequency of word i appearing in a specific document j; is the weight of word i appearing in a specific document j (the fewer the number of documents containing word i, the greater the weight), and N is the total number of documents in the corpus.

[0060] The specific document j may be a document in Excel format, text format, etc. (if the document is in Word or PDF format, it may be processed first to be converted into Excel or text format), and

[0061] Step S203, filtering out words whose importance factors are greater than or equal to the importance factor threshold from the multiple words, and adding the filtered words to the first keyword group.

[0062] If the importance factor of a word is greater than the importance factor threshold, it indicates that the word is highly correlated with the change operation that leads to the preset change consequence. The word needs to be screened out and added to the first keyword group for use in the subsequent keyword recommendation process for evaluating change risks.

[0063] After executing step S102, the establishment method may also include: classifying the multiple first keywords according to the characteristics of the multiple first keywords in the first keyword group to obtain the multiple first keywords of different classifications; and constructing a change keyword library with the multiple first keywords of different classifications.

[0064] Specifically, firstly, the words in the above-mentioned first keyword group are deduplicated based on rule matching, and then the characteristics of the keywords in the first keyword group are divided into four categories of keywords related to change devices, change descriptions, change risks, and control measures, and the four categories of keywords are further subjected to feature analysis to be divided into three levels, thereby constructing a three-level change keyword library (see Table 2, where the third-level keywords are specific descriptions, which are not reflected in Table 2).

[0065] That is to say, Jieba word segmentation method and TF-IDF weighting method can be executed through steps S201-S203, and after matching and removing duplicates based on the change instance using preset rules, four types of keywords related to the change device, change description, change risk and control measures can be extracted.

[0066] Table 2 Description of the change keyword library classification

[0067]

[0068] The establishing of the keyword recommendation model may include: constructing the keyword recommendation model in the form of a graph database based on a plurality of first specific keywords of different categories with preset associations.

[0069] A graphical database can be used to establish an internal and inter-class impact association model (i.e., a keyword recommendation model) for four types of keywords, including change description (process card change), change risk (failure to inform the operator, untrained operator, display error, operator unfamiliarity, misoperation, equipment damage), and control measures (enhanced inspection, timely notification, etc.). Figure 3 shown.

[0070] After obtaining the keyword recommendation model through step S103, the keyword recommendation model can be trained using historical data, so that when a new change request is introduced, keywords corresponding to the change request can be recommended based on the trained recommendation model (the keywords can be used to evaluate the change risk corresponding to the change request).

[0071] To sum up, the present invention creatively identifies multiple risk factors related to change operations that lead to specific change consequences comprehensively and objectively based on change instances and change accidents, then extracts multiple keywords based on the identified risk factors, and then establishes a corresponding keyword recommendation model based on multiple specific keywords with certain correlations among the extracted multiple keywords. As a result, keywords for evaluating change risks can be intelligently recommended based on the established keyword recommendation model, thereby improving the accuracy and efficiency of change risk assessment, and further realizing intelligent change risk management.

[0072] Embodiment 3

[0073] Figure 4 is a flow chart of a method for recommending keywords for evaluating change risk provided by Embodiment 3 of the present invention. Figure 4 As shown, the recommendation method may include: step S401, determining the similarity between the change request and each of the multiple keyword recommendation models obtained according to the method for establishing the keyword recommendation model; and step S402, when the similarity between the change request and a specific keyword recommendation model among the multiple keyword recommendation models is greater than or equal to a preset similarity threshold, the specific keyword recommendation model recommends a plurality of first specific keywords corresponding to the change request.

[0074] Specifically, if the similarity between the change request and the specific keyword recommendation model is greater than or equal to the preset similarity threshold, it indicates that the change request is strongly associated with multiple keywords included in the specific keyword recommendation, and the specific keyword recommendation model outputs the corresponding multiple keywords, thereby performing a risk assessment of the change request based on the output multiple keywords.

[0075] Embodiment 4

[0076] On the basis of the above-mentioned embodiment three, if it is determined that there is no strong correlation between the change request and any keyword recommendation model, it is necessary to match and push the change request based on the keyword group extracted from the change accidents in the change accident case library and the change instances in the change instance library by taking the intersection of the change request and the keyword group.

[0077] The recommendation method may also include: when the similarity between the change request and any one of the multiple keyword recommendation models is less than the preset similarity threshold, matching the change request with multiple second keywords in the second keyword group in the change keyword library to obtain multiple second specific keywords corresponding to the change request, wherein the multiple second keywords are extracted based on the change accident and the change instance; and recommending multiple second specific keywords corresponding to the change request.

[0078] Specifically, the change accident and the change instance may be processed in advance with word segmentation, weighting and feature analysis to obtain a second keyword group corresponding to the change accident and the change instance, wherein the second keyword group includes a plurality of second keywords; and then the second keyword group is added to the above-mentioned change keyword library. Here, the process of performing word segmentation, weighting and feature analysis can be found in the specific description of the above-mentioned steps S201-S203, which will not be repeated here. On this basis, the change request is intersected with the plurality of second keywords in the second keyword group to match a plurality of specific keywords; and then the matched plurality of specific keywords are recommended.

[0079] Therefore, two recommendation rules can be formulated according to the strength of the association between change requests and keywords, so that risks, possible consequences, control measures and other content can be automatically associated when adding new change requests.

[0080] Embodiment 5

[0081] On the basis of the above-mentioned embodiment three or embodiment four, the recommendation priority of the multiple keyword settings can also be set according to the strength of the conditional dependency relationship between the multiple keywords and the change request, thereby giving priority to the recommendation of keywords with stronger conditional dependency relationships. Specifically, in the case where keywords such as new, demolish, cancel, adjust, unreasonable or rectify are stored in the change keyword library, when a "new" request appears, after the conditional dependency analysis, since the dependency on the change objects such as "pipeline" and "blind plate" is stronger, "new" is preferentially related to "pipeline" and "blind plate", that is, "pipeline", "blind plate" and the like are preferentially recommended, thereby leading to subsequent associated risks. In addition, since the conditional dependency relationship between the keywords with a lower recommendation priority and the change request is weaker (the keywords with a lower recommendation priority can be ignored), it can also be set to output only the keywords at the first few recommendation priorities according to the actual situation.

[0082] The recommendation method may further include: determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords based on the conditional dependency relationship between the recommended multiple first specific keywords or the multiple second specific keywords and the change request, and accordingly, recommending the multiple first specific keywords or the multiple second specific keywords corresponding to the change request includes: according to the recommendation priority of the multiple first specific keywords or the multiple second specific keywords, screening and recommending the first specific keywords or the second specific keywords with a recommendation priority ranked in the front preset ranking.

[0083] Specifically, determining the recommendation priority of the multiple first keywords or the multiple second keywords may include: calculating the conditional probability of each of the output multiple first specific keywords or the multiple second specific keywords being recommended when the change request occurs based on a conditional probability formula; and determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords according to the conditional probability of each of the multiple first specific keywords or the multiple second specific keywords being recommended when the change request occurs.

[0084] Specifically, for the change request W, the recommended keyword C is used as an example for explanation.

[0085] According to the conditional probability formula P(C|W)=P(W|C)*P(C) / P(W) (where P(W|C) represents the probability of change request W being established when keyword C appears; P(C) represents the probability of keyword C appearing; and P(W) represents the probability of change request W being established), it can be seen that for all candidate keywords C, the corresponding P(W) is the same, so P(C|W) is proportional to P(W|C)*P(C). When the change database is large enough, P(C) is the frequency of keyword C appearing; and P(W|C) can be obtained by statistical probability (i.e. frequency). Specifically, in the change instance library or the change accident case library, the number of entries containing keyword C and W is divided by the number of entries containing keyword C (which can be retrieved and matched by regular expression). For example, searching for "W and C" matches 20 results, and searching for C alone produces 50 results, then P(W|C)=2 / 5.

[0086] Therefore, when P(C|W) is the largest, when a change request W is input, the keyword C most relevant to the change request W is inferred, so that keyword C can be recommended first, and other recommended contents can be sorted in sequence and recommended later.

[0087] When the recommendation priority of each specific keyword is obtained, only the top N (for example, top 3) priority specific keywords may be recommended in the order of recommendation priority.

[0088] Embodiment 6

[0089] For step S401, Figure 5 As shown, the determining of the similarity between the change request and each of the plurality of keyword recommendation models obtained according to the keyword recommendation model establishment method may include steps S501-S503.

[0090] Step S501: determine the vector corresponding to the change request.

[0091] The elements in the vector (i.e., one-dimensional vector) can be arranged in a preset word order, and the value of each element in the vector is the number of times each corresponding word appears repeatedly. Thus, the change request can be segmented to obtain multiple words, and the vector corresponding to the change request can be determined according to the specific conditions of the multiple words obtained.

[0092] If the elements in the vector are arranged in the order of the pinyin of the words, when the change request is a process card change, the corresponding vector is [1 (change) 1 (process) 1 (card)].

[0093] Step S502: determining a vector corresponding to each of the plurality of keyword recommendation models.

[0094] Accordingly, the vector corresponding to each keyword recommendation model may be determined according to the above-mentioned preset word order, which will not be described in detail here.

[0095] Step S503 : determining the similarity between the change request and each of the multiple keyword recommendation models according to the cosine similarity between the vector corresponding to the change request and the vector corresponding to each of the multiple keyword recommendation models.

[0096] The larger the cosine similarity between the vector corresponding to the change request and the vector corresponding to the keyword recommendation model, the higher the similarity between the change request and the keyword recommendation model. Therefore, when the similarity between the change request and the specific keyword recommendation model is greater than or equal to the preset similarity threshold, it indicates that the multiple keywords output by the specific keyword recommendation model can be used to evaluate the change risk of the change request.

[0097] To sum up, the present invention creatively determines the similarity between the change request and each of the multiple keyword recommendation models obtained according to the above-mentioned keyword recommendation model establishment method. If the similarity between the change request and the specific keyword recommendation model is greater than or equal to the preset similarity threshold, the specific keyword recommendation model outputs the corresponding multiple specific keywords. Therefore, the change request can be intelligently risk evaluated based on the output multiple keywords, thereby improving the accuracy and efficiency of the change risk assessment, and further realizing intelligent change risk management.

[0098] Embodiment 7

[0099] In the seventh embodiment of the present invention, the recommendation process of keywords for evaluating change risk is explained and illustrated.

[0100] like Figure 6 As shown, the keyword recommendation process for evaluating change risk includes steps S601-S614.

[0101] Step S601, obtaining a change instance and a change accident.

[0102] Afterwards, step S602 and step S608 are executed.

[0103] Step S602, based on the evolutionary time sequence relationship and evolution law of the change accident and the change instance, infer the uncontrolled link 1, uncontrolled link 2... and uncontrolled link N of the risk evolution of the preset change consequence.

[0104] Step S603: Based on the change instance, a logic tree analysis method is used to identify risk factor 1, risk factor 2, ..., and risk factor M associated with the inferred out-of-control link.

[0105] Step S604: perform word segmentation, weighting and feature analysis based on multiple risk factors to obtain a first keyword group.

[0106] Step S605: adding the first keyword group to the changed keyword library.

[0107] Step S606: establishing a keyword recommendation model based on a plurality of specific keywords with preset relevance in the first keyword group.

[0108] Step S607: adding the keyword recommendation model to the model library.

[0109] Afterwards, execute step S610.

[0110] Step S608: performing word segmentation, weighting and feature analysis on the change instance and the change accident to obtain a second keyword group.

[0111] Step S609: adding the second keyword group to the changed keyword library.

[0112] Afterwards, execute step S610.

[0113] Step S610, determining whether the change request has a strong correlation with the specific keyword recommendation model in the model library, if so, executing step S611; otherwise, executing step S612.

[0114] Step S611: the specific keyword recommendation model outputs a plurality of specific keywords corresponding to the change request.

[0115] Step S612: Matching with the keywords in the second specific keyword group to output a plurality of specific keywords corresponding to the change request.

[0116] Step S613: determining the recommendation priorities of the multiple specific keywords based on the conditional dependency relationship.

[0117] Step S614: recommend the multiple specific keywords according to the determined recommendation priorities.

[0118] Therefore, through the two recommendation methods in this embodiment, keywords associated with the change request can be automatically recommended. For example, the keywords may include risk factors, possible consequences and control measures, etc., so that the risks that may be brought about by the above change request can be effectively evaluated.

[0119] Embodiment 8

[0120] Figure 7 FIG. 8 is a structural diagram of a system for establishing a keyword recommendation model provided by Embodiment 8 of the present invention. Figure 7 As shown, the establishment system may include: a risk factor identification device 10, used to identify multiple risk factors related to the change operation that leads to preset change consequences based on change instances and change accidents; a keyword extraction device 20, used to extract a first keyword group from the multiple risk factors, the first keyword group including multiple first keywords; and a model establishment device 30, used to establish the keyword recommendation model based on multiple first specific keywords with preset correlation in the first keyword group.

[0121] Preferably, the risk factor identification device 10 includes: an inference module (not shown), which is used to infer the uncontrolled link of the risk evolution of the preset change consequence based on the evolutionary time series relationship and evolution law of the change accident and the change instance; and a risk factor identification module (not shown), which is used to identify the multiple risk factors associated with the inferred uncontrolled link based on the change instance by using a logic tree analysis method.

[0122] Preferably, the keyword extraction device 20 includes: a word segmentation module (not shown), used to segment the multiple risk factors to obtain multiple words; a factor calculation module (not shown), used to calculate the importance factor of each of the multiple words in a specific document in a corpus, wherein the corpus is formed by the change instance and the change accident; and a screening module (not shown), used to screen out words whose importance factor is greater than or equal to the importance factor threshold from the multiple words, and add the screened words to the first keyword group.

[0123] Preferably, the establishment system also includes: a classification device (not shown) for classifying the multiple first keywords according to the characteristics of the multiple first keywords in the first keyword group to obtain the multiple first keywords of different classifications; and a construction device (not shown) for constructing a change keyword library with the multiple first keywords of different classifications.

[0124] The risk factor identification device 10, the keyword extraction device 20, the classification device (not shown) and the construction device (not shown) can be integrated in the same device (e.g. Figure 8 The keyword library construction device 40 can be connected to the model building device 30 and can provide a construction object for the model building device 30, such as Figure 8 shown.

[0125] The specific details and benefits of the system for establishing the keyword recommendation model provided by the present invention can be found in the description of the method for establishing the keyword recommendation model mentioned above, which will not be repeated here.

[0126] Embodiment 9

[0127] Fig. 9 It is a structural diagram of a keyword recommendation system for evaluating change risk provided by Embodiment 9 of the present invention. The recommendation system comprises: a similarity determination device 50, for determining the similarity between the change request and each of the multiple keyword recommendation models acquired by the establishment system of the keyword recommendation model; and a first recommendation device 60, for recommending multiple first keywords corresponding to the change request by the specific keyword recommendation model when the similarity between the change request and a specific keyword recommendation model among the multiple keyword recommendation models is greater than or equal to a preset similarity threshold.

[0128] Preferably, the recommendation system also includes: a matching device (not shown) for matching the change request with a plurality of second keywords in a second keyword group in a change keyword library when the similarity between the change request and any one of the plurality of keyword recommendation models is less than the preset similarity threshold, so as to obtain a plurality of second specific keywords corresponding to the change request, wherein the plurality of second keywords are extracted based on the change instance; and a second recommendation device (not shown) for recommending a plurality of second specific keywords corresponding to the change request.

[0129] The first recommendation device 60 and the second recommendation device (not shown) output various risks and control measures that may be brought about by the change request, so as to realize active prompting of potential change risks. The first recommendation device 60 can be set to the following output mode: after the assessor enters the change request (for example, the change vocabulary), a strong association recommendation is made based on the input change request, and data with strong conditional dependence is pushed first. The second recommendation device (not shown) can be set to the following output mode: click the check box to recommend the subject word (equivalent to inputting the change request), and call the sample from the instance library to realize the association risk recommendation.

[0130] Preferably, the recommendation system further comprises: a priority determination device 70, which can be connected to the keyword library construction device 40 (such as Figure 8), for determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords based on the conditional dependency relationship between the recommended multiple first specific keywords or the multiple second specific keywords and the change request. Accordingly, the first recommendation device 60 is used to recommend the multiple first specific keywords corresponding to the change request or the second recommendation device (not shown) is used to recommend the multiple second specific keywords corresponding to the change request, including: according to the recommendation priority of the multiple first specific keywords or the multiple second specific keywords, screening and recommending the first specific keywords or the second specific keywords with the recommendation priority ranked in the front preset ranking.

[0131] The priority determination device (not shown) includes: a probability calculation module (not shown) for calculating, based on a conditional probability formula, the conditional probability that each of the outputted multiple first specific keywords or the multiple second specific keywords is recommended when the change request occurs; and a priority determination module (not shown) for determining the recommendation priority of the multiple first specific keywords or the multiple second specific keywords according to the conditional probability that each of the outputted multiple first specific keywords or the multiple second specific keywords is recommended when the change request occurs.

[0132] Preferably, the similarity determination device 50 includes: a first vector determination module (not shown), used to determine the vector corresponding to the change request; a second vector determination module (not shown), used to determine the vector corresponding to each of the multiple keyword recommendation models; and a similarity determination module (not shown), used to determine the similarity between the change request and each of the multiple keyword recommendation models based on the cosine similarity between the vector corresponding to the change request and the vector corresponding to each of the multiple keyword recommendation models.

[0133] The specific details and benefits of the keyword recommendation system for evaluating change risk provided by the present invention can be found in the description of the keyword recommendation method for evaluating change risk, which will not be repeated here.

[0134] Another embodiment of the present invention further provides a machine-readable storage medium, on which instructions are stored, and the instructions are used to enable a machine to execute the method for establishing the keyword recommendation model and / or the method for recommending keywords for evaluating change risks.

[0135] The machine-readable storage medium includes but is not limited to phase change memory (short for phase change random access memory, Phase Change Random Access Memory, PRAM, also known as RCM / PCRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory (Flash Memory) or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices, and other media that can store program codes.

[0136] The preferred embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, a variety of simple modifications can be made to the technical solution of the present invention, and these simple modifications all belong to the protection scope of the present invention.

[0137] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present invention will not further describe various possible combinations.

[0138] In addition, various embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the present invention, they should also be regarded as the contents disclosed by the present invention.

Claims

1. A method for establishing a keyword recommendation model, characterized in that: The establishment method comprises: Based on change instances and change incidents, identify multiple risk factors associated with change operations that lead to pre-set change consequences; Extracting a first keyword group from the plurality of risk factors, the first keyword group including a plurality of first keywords; and Based on a plurality of first specific keywords with preset associations in the first keyword group, the keyword recommendation model is established. The identifying of multiple risk factors associated with the change operation leading to the preset change consequence includes: Based on the evolutionary time sequence relationship and evolutionary law of the change accident and the change instance, infer the uncontrolled link of the risk evolution of the preset change consequence; and Based on the change instance, a logic tree analysis method is used to identify the multiple risk factors associated with the inferred out-of-control link. The step of establishing the keyword recommendation model includes: constructing the keyword recommendation model in the form of a graphic database based on a plurality of first specific keywords of different categories with preset associations.

2. The method for establishing a keyword recommendation model according to claim 1, characterized in that: The extracting the first keyword group from the multiple risk factors comprises: Segmenting the multiple risk factors to obtain multiple words; Calculating an importance factor of each of the plurality of words in a specific document in a corpus, wherein the corpus is formed by the change instances and the change incidents, and the specific document is a document in an Excel format or a text format; and The words whose importance factors are greater than or equal to the importance factor threshold are screened out from the multiple words, and the screened words are added to the first keyword group.

3. The method for establishing a keyword recommendation model according to claim 2, characterized in that: The calculating the importance factor of each of the plurality of words in a specific document in the corpus comprises: The importance factor of each of the plurality of words in a specific document in the corpus is calculated using the following weighting algorithm: W tf (i,j)=tf(i,j)×idf(i), Where tf(i, j) is the frequency of word i appearing in a specific document j; is the weight of word i appearing in a specific document j, and N is the total number of documents in the corpus.

4. The method for establishing a keyword recommendation model according to claim 1, characterized in that: After executing the step of extracting the first keyword group from the multiple risk factors, the establishment method further includes: classifying the plurality of first keywords according to the characteristics of the plurality of first keywords in the first keyword group to obtain the plurality of first keywords of different classifications; and A changed keyword library including the plurality of first keywords of different classifications is constructed.

5. A method for recommending keywords for evaluating change risk, characterized in that: The recommended methods include: Determine the similarity between the change request and each of the plurality of keyword recommendation models obtained according to the method for establishing a keyword recommendation model according to any one of claims 1 to 4; and When the similarity between the change request and a specific keyword recommendation model among the multiple keyword recommendation models is greater than or equal to a preset similarity threshold, the specific keyword recommendation model recommends multiple first specific keywords corresponding to the change request.

6. The method for recommending keywords for evaluating change risk according to claim 5, characterized in that: The recommended method also includes: In a case where the similarity between the change request and any one of the plurality of keyword recommendation models is less than the preset similarity threshold, matching the change request with a plurality of second keywords in a second keyword group in a change keyword library to obtain a plurality of second specific keywords corresponding to the change request, wherein the plurality of second keywords are extracted based on the change instance; and A plurality of second specific keywords corresponding to the change request are recommended.

7. The method for recommending keywords for evaluating change risk according to claim 5 or 6, characterized in that: The recommended method also includes: determining a recommendation priority of the plurality of first specific keywords or the plurality of second specific keywords based on a conditional dependency relationship between the recommended plurality of first specific keywords or the plurality of second specific keywords and the change request, Accordingly, the recommending a plurality of first specific keywords or a plurality of second specific keywords corresponding to the change request includes: According to the recommendation priorities of the plurality of first specific keywords or the plurality of second specific keywords, the first specific keyword or the second specific keyword with a recommendation priority at a front preset ranking is screened and recommended.

8. The method for recommending keywords for evaluating change risk according to claim 7, characterized in that: The determining of the recommendation priorities of the plurality of first specific keywords or the plurality of second specific keywords comprises: Based on a conditional probability formula, calculating a conditional probability that each of the outputted plurality of first specific keywords or the plurality of second specific keywords is recommended when the change request occurs; and The recommendation priority of the plurality of first specific keywords or the plurality of second specific keywords is determined according to a conditional probability that each of the plurality of first specific keywords or the plurality of second specific keywords is recommended when the change request occurs.

9. The method for recommending keywords for evaluating change risk according to claim 5, characterized in that: The determining of the similarity between the change request and each of the plurality of keyword recommendation models obtained according to the keyword recommendation model establishment method comprises: Determining a vector corresponding to the change request; Determining a vector corresponding to each of the plurality of keyword recommendation models; and The similarity between the change request and each of the plurality of keyword recommendation models is determined according to the cosine similarity between the vector corresponding to the change request and the vector corresponding to each of the plurality of keyword recommendation models.

10. A system for establishing a keyword recommendation model, characterized in that: The establishment system comprises: A risk factor identification device, for identifying a plurality of risk factors associated with a change operation leading to a preset change consequence based on a change instance and a change accident; A keyword extraction device, configured to extract a first keyword group from the plurality of risk factors, the first keyword group including a plurality of first keywords; and A model building device is used to build the keyword recommendation model based on a plurality of first specific keywords with preset associations in the first keyword group, The identifying of multiple risk factors associated with the change operation leading to the preset change consequence includes: Based on the evolutionary time sequence relationship and evolutionary law of the change accident and the change instance, infer the uncontrolled link of the risk evolution of the preset change consequence; and Based on the change instance, a logic tree analysis method is used to identify the multiple risk factors associated with the inferred out-of-control link. The step of establishing the keyword recommendation model includes: constructing the keyword recommendation model in the form of a graphic database based on a plurality of first specific keywords of different categories with preset associations.

11. A keyword recommendation system for evaluating change risk, characterized in that: The recommendation system comprises: a similarity determination device for determining a similarity between the change request and each of the plurality of keyword recommendation models acquired by the keyword recommendation model establishment system according to claim 10; and A keyword output device is used to recommend multiple first specific keywords corresponding to the change request by the specific keyword recommendation model when the similarity between the change request and a specific keyword recommendation model among the multiple keyword recommendation models is greater than or equal to a preset similarity threshold.

12. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores instructions for causing a machine to execute the method for establishing a keyword recommendation model according to any one of claims 1 to 4 and / or the method for recommending keywords for assessing change risks according to any one of claims 5 to 9.

Images