A method and system for quickly extracting three-cause rules of traffic accidents

Abstract

The present invention discloses a method and system for rapidly extracting three-branch causal rules of traffic accidents, comprising the following steps: constructing a three-branch concept lattice derived from an object using an incremental construction algorithm; extracting three-branch decision association rules based on the three-branch concept lattice derived from the object; and rapidly extracting three-branch causal rules of traffic accidents. The method of the present invention is applied to mining traffic accident causal rules, thereby obtaining representative three-branch rules of traffic accident causation. After analysis, it is found that the three-branch decision association rules are more semantically richer and more delicately expressed than the two-branch decision association rules, and can express knowledge that the two-branch decision rules cannot express, and can mine richer and more comprehensive traffic accident mechanisms.

Description

Technical Field

[0001] The present invention relates to the field of three-cause concept analysis, and in particular to a method and system for quickly extracting three-cause rules of traffic accidents. Background Art

[0002] Formal concept analysis (FCA) is an effective mathematical tool for conceptual knowledge representation and discovery. In classic FCA, a two-way decision strategy (acceptance and rejection) is adopted. In fact, three-way decision is more suitable for dealing with decision problems in the real world. Three-way concept analysis is proposed by combining the ideas of FCA and three-way decision, and is an extension of classic formal concept analysis. At present, extracting three-way decision rules based on OE3W concept lattice is a hot research topic in three-way concept analysis, which helps three-way concept analysis solve complex decision problems in practical applications and improve decision efficiency.

[0003] There are many methods for analyzing the causes of traffic accidents, and association rule analysis is a commonly used data mining research method in traffic accident cause analysis. The current association rule analysis algorithms mainly include FP-Tree (Frequent Pattern Tree) algorithm and Apriori algorithm. The Apriori algorithm can mine the correlation between two or more factors, but when the data volume is large, the computational efficiency will be significantly reduced. In order to improve the accuracy and efficiency of the Apriori algorithm, an improved Apriori algorithm was proposed, but it is lacking in the degree of knowledge expression. FP-Tree (Frequent Pattern Tree) is also an association rule data mining algorithm, but the construction of FP-Tree requires multiple scans and sorting of the data set, which increases the construction time and computational cost, especially when the data volume is large. It can be found that the traditional traffic accident cause rule is a two-branch decision rule, which is lacking in rule extraction efficiency and the degree of knowledge expression. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a method and system for quickly extracting three-branch causal rules of traffic accidents, a method for extracting three-branch decision association rules (representative three-branch decision association rules) based on an object-derived concept lattice (referred to as OE3W concept lattice), and applies this method to the mining of three-branch causal rules of traffic accidents, solving the problems that traditional traffic accident causal rules (two-branch decision rules) are lacking in rule extraction efficiency and the degree of knowledge expression.

[0005] For the purpose of describing the present invention, some symbols are defined. Given a decision formal background T = (G, A, I, D, J), assume that the formal background T c =(G,A,I) generated by OE3W concept lattice is recorded as OE3WL(T c ). Let (G,A,I) be the formal background. For any OE3W concept (H, (O, S)), OE3W concept (P, (R, T)), then *, The operator and the partial order relation ≥ are defined as:

[0006]

[0007] Where I = G × A.

[0008] The incremental algorithm constructs the OE3W concept lattice by incrementally adding objects in the formal background to the formal background. c ) and OE3WL'(T c ) represent the OE3W concept grids corresponding to the formal background before and after the newly added object g. In order to indicate that the formal background has changed, we will use and to represent the concept derivation operators in the formal background before and after the newly added object g, ≥ and ≥' respectively represent the partial order relationship between the concepts before and after the newly added object, and at the same time, it is assumed that the attribute set in the background will not change.

[0009] To achieve the above object, the present invention provides the following technical solution: a method for quickly extracting the three-cause rules of traffic accidents, comprising the following steps:

[0010] S1, construct the OE3W concept lattice using incremental algorithm;

[0011] S2, extracting three-branch decision association rules based on OE3W concept lattice;

[0012] S3. Rapid extraction of the three causal rules of traffic accidents.

[0013] Preferably, in step S1, the following is specifically included:

[0014] S11, add the OE3WL'(T c ) are divided into three categories: new OE3W concept, revised OE3W concept, and old OE3W concept;

[0015] S12. Identify the necessary and sufficient conditions for the three types of OE3W concepts: new OE3W concepts, revised OE3W concepts, and old OE3W concepts;

[0016] S13. Update the order relation. For each new OE3W concept (H'0, (O'0, S'0)), its standard generator is (H0, (O0, S0)), which must be the new OE3W concept (H' 0, The lower nearest neighbor of (O'0,S'0);

[0017] For each new OE3W concept (H'0, (O'0, S'0)), determining its neighbor set is divided into two steps:

[0018] 1) Determine the candidate upper nearest neighbor set C = C1∪C2, where C1 and C2 are:

[0019]

[0020] 2) The minimum OE3W concept set C' in the candidate upper nearest neighbor set C is the upper nearest neighbor set of the new OE3W concept (H'0, (O'0, S'0));

[0021] For the modified OE3W concept in the upper neighbor set C', the order relationship between the modified OE3W concept and its child concepts and the parent concept needs to be deleted. According to the above process, the order relationship is updated;

[0022] S14. Through steps S11-S13, all OE3W concepts after the newly added object g can be updated, that is, all modified OE3W concepts, new OE3W concepts and the update of the order relationship after the added object g are found; continue to add the object g that has not been added in G, repeat steps S11-S14, traverse all objects in G, and finally obtain the conditional form sub-context T c OE3W concept grid OE3WL (T c ).

[0023] Preferably, in step S11, the following is specifically included:

[0024] S111、If (H', (O', S'))∈OE3WL'(T c ), and its connotation (O',S') is not OE3WL(T c ), then (H', (O', S')) is called a new OE3W concept;

[0025] S112、If (H, (O, S))∈OE3WL(T c ), (H∪g,(O,S))∈OE3WL'(T c ), then (H, (O, S)) is called the modified OE3W concept;

[0026] S113、If (H, (O, S))∈OE3WL(T c ), and (H,(O,S))∈OE3WL'(T c ), then (H,(O,S)) is called the old OE3W concept.

[0027] Preferably, in step S12, the following is specifically included:

[0028] S121、If (H, (O, S))∈OE3WL(T c ), (H', (O', S'))∈OE3WL'(T c ), Then (H', (O', S')) is called a new OE3W concept. In order to speed up the generation of new OE3W concepts, a standard generator is introduced. For any new OE3W concept (H', (O', S')), is the standard generator of (H',(O',S'));

[0029] S122、If (H, (O, S))∈OE3WL(T c ), Then it means that (H, (O, S)) is a modified OE3W concept;

[0030] S123、If (H, (O, S))∈OE3WL(T c ), (H', (O', S'))∈OE3WL'(T c ), Then (H, (O, S)) is called the old OE3W concept.

[0031] Preferably, in step S2, the following is specifically included:

[0032] S21. Definition of three-branch decision association rule: Let T = (G, A, I, D, J) be a decision form background. For any X, in if θ, ρ are thresholds set in advance, then are three-branch decision association rules, and their support and confidence are recorded as

[0033] S22. Distinguish decision association rules into representative three-branch decision association rules and non-representative three-branch decision association rules: If are two three-way decision association rules of T, if and but Can be Indicates that For any If not present Then it is called is a representative three-branch decision association rule, otherwise it is called is a non-representative three-way decision association rule;

[0034] S23. Extraction of representative three-branch decision association rules: First, use the support threshold θ to identify OE3WL(T c ) to prune, and the pruned OE3WL (T c ) is recorded as OE3WL * (T c ); for OE3WL * (T c ) in any OE3W concept (H, (O, S)), for any attribute Z in D, if it satisfies Then Add to the set of representative three-branch decision association rules Ψ(T); traverse each attribute in D and OE3WL * (T c ) for each OE3W concept, and finally obtain all the representative three-way decision association rules Ψ(T);

[0035] S24. Extraction of non-representative three-branch decision association rules: For any representative three-branch decision association rule First, find the set of all intensional approximations of (O,S) ((X,P) is called intensional approximation of (O,S) if and only if X * ∩P°=O * ∩S°=H), if (X,P) is a pseudo-intension of (O,S) (when Then (X,P) is a pseudo-intension of (O,S)), then Add the set of non-representative three-way decision association rules Φ(T); traverse the representative three-way decision association rules in Ψ(T), and finally obtain all non-representative three-way decision association rules Φ(T).

[0036] Preferably, in step S24, for obtaining the intension approximate set of any OE3W concept (H, (O, S)), the specific steps are as follows: TEMP is a set of subsets of the attribute set (O, S), for any attribute set (X, P) in TEMP, if there is no parent concept (H'0, (O'0, S'0)) of (H, (O, S)), such that Then add (X,P) to the connotation approximation set of the OE3W concept (H,(O,S)).

[0037] Preferably, in step S3, the following is specifically included:

[0038] Firstly, the attributes of the driver, vehicle, road and environment dimensions in traffic accidents are taken as conditional attributes, which are processed as conditional formal sub-contexts, and the severity classification of the accident is taken as the decision attribute; secondly, the obtained conditional formal sub-context of traffic accidents is gridded to obtain the three-branch concept lattice derived from the object, and the representative three-branch causal rules of traffic accidents are extracted using the OE3W concept lattice.

[0039] Through mathematical proof, we can find the relationship between the three-way decision association rule and the two-way decision association rule:

[0040] 1. For any two-branch decision association rule O→Z, we can find the only one that satisfies Three-branch decision association rule Where S = O * °.

[0041] 2. For any three-branch decision association rule (1) When S = O * °, there is a satisfying The two-branch decision association rule O→Z; (2) when There is a satisfying The two-branch decision association rule O→Z; (3) when S≠O * °, There is no two-branch decision association rule O→Z.

[0042] The three-way decision association rule is a supplement to the two-way decision association rule and can express knowledge that the two-way rule cannot express. Therefore, applying the three-way decision association rule to the mining of traffic accident causal rules can mine a richer and more comprehensive traffic accident causal mechanism.

[0043] On the other hand, in order to achieve the above-mentioned purpose, the present invention also provides the following technical solution: a rapid extraction system of three-cause rules of traffic accidents, the system comprising the following modules:

[0044] The incremental algorithm constructs the OE3W concept lattice module (110), updates the three concepts derived from the object and the order relationship by analyzing the relationship between the three concept lattices derived from the objects before and after the newly added object g;

[0045] Extracting three-branch decision association rule module (120), firstly giving a definition of three-branch decision association rule, secondly distinguishing decision association rule into representative and non-representative three-branch decision association rule, and then giving a method for extracting representative and non-representative three-branch decision association rule based on OE3W concept lattice;

[0046] The module (130) for quickly extracting the three-cause rules of traffic accidents takes the attributes of the driver, vehicle, road and environment dimensions in the traffic accident as conditional attributes, processes them as conditional form sub-backgrounds, and classifies the severity of the accident as decision attributes; then constructs a grid based on the obtained traffic accident conditional form sub-backgrounds to obtain the three-branch concept grid derived from the object, and uses the OE3W concept grid to extract representative three-cause rules of traffic accidents.

[0047] The beneficial effects of the present invention are:

[0048] 1) The present invention adopts an incremental algorithm to construct the OE3W concept lattice, and extracts three-branch decision association rules on this basis. The method of extracting three-branch decision association rules based on the OE3W concept lattice is applied to the mining of traffic accident causal rules, which can effectively improve the efficiency of mining three-branch causal rules of traffic accidents.

[0049] 2) The three-branch decision association rule based on the OE3W concept lattice proposed in the present invention has richer semantics and more refined expression than the two-branch decision association rule, and can express knowledge that the two-branch decision rule cannot express. Therefore, the method of extracting three-branch decision association rules based on the OE3W concept lattice is applied to the mining of traffic accident causal rules, which can mine a richer and more comprehensive traffic accident causal mechanism than the two-branch causal rule. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Schematic diagram of a method for quickly extracting three-cause rules of traffic accidents in an embodiment of the present invention;

[0051] Figure 2 It is an updated diagram of the OE3W concept in OE3WL(T) after adding the object g in the embodiment of the present invention;

[0052] Figure 3 Schematic diagram of the OE3W concept lattice obtained from the conditional form sub-context (including objects 1, 2, and 3) in Table 1 in an embodiment of the present invention;

[0053] Figure 4 Schematic diagram of the OE3W concept lattice obtained from the conditional form sub-context (including objects 1, 2, 3, and 4) in Table 1 in an embodiment of the present invention;

[0054] Figure 5 It is a flowchart of the incremental construction algorithm - the process of establishing the order relationship between OE3W concepts in an embodiment of the present invention;

[0055] Figure 6 Flow chart of extracting three-branch decision association rules based on OE3W concept lattice in an embodiment of the present invention;

[0056] Figure 7Schematic diagram of a system module for quickly extracting three-cause rules of traffic accidents in an embodiment of the present invention;

[0057] In the figure, 110-incremental algorithm constructs OE3W concept lattice module; 120-extracts three-branch decision association rule module; 130-traffic accident three-branch causal rule fast extraction module. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] The present invention provides a technical solution: a method for quickly extracting the three-cause rule of traffic accidents, the steps are as follows: Figure 1 As shown, specifically including:

[0060] 1. Use incremental algorithm to build OE3W concept lattice

[0061] The incremental algorithm constructs the OE3W concept lattice by incrementally adding objects in the formal background to the formal background. c ) and OE3WL'(T c ) represent the OE3W concept lattice corresponding to the formal background before and after the newly added object g, and the process is as follows Figure 2 shown.

[0062] Given a decision-making context T = (G, A, I, D, J) (Table 1). c =(G,A,I) is the conditional form sub-context of Table 1, such as Figure 3 As shown, the conditional form of Table 1 is constructed by the incremental construction method of the OE3W concept lattice proposed by the present invention. c The OE3W concept grid.

[0063] Table 1 Decision-making form background T

[0064]

[0065] In the formal context of objects 1, 2, and 3 (the constructed OE3W concept grid is as follows Figure 3 ), add object 4 to the formal background, and get the updated OE3W concept lattice OE3WL'(T c ),like Figure 4 As shown, the new OE3W concept has (124,(ac,b)), (24,(ac,be)), (4,(ac,bde)).

[0066] For the update of the order relation of the new OE3W concept (24, (ac, be)) (which is essentially to find the upper and lower neighbors of the new OE3W concept (24, (ac, be)), such as Figure 5 As shown, specifically including:

[0067] For the new OE3W concept (24, (ac, be)), its standard generator is (2, (acd, be)), which must be the lower neighbor of the new OE3W concept (24, (ac, be)).

[0068] For the new OE3W concept (24, (ac, be)), determining its neighbor set is divided into two steps:

[0069] (1) Determine the candidate upper nearest neighbor set C = C1 ∪ C2, where

[0070]

[0071] The minimum OE3W concept set C'=(124,(ac,b)) in the candidate upper neighbor set C is the lower neighbor set of the new OE3W concept (24,(ac,be)).

[0072] According to the above idea, the order relationship of the new OE3W concept (24, (ac, be)) is updated.

[0073] 2. Extracting three-branch decision association rules based on OE3W concept lattice

[0074] The method proposed in this invention for extracting three-branch decision association rules based on OE3W concept lattice is as follows: Figure 6 As shown, based on the OE3W concept lattice constructed above and the given decision form background T = (G, A, I, D, J) (Table 1), three decision association rules are extracted, and the given support threshold (θ) and confidence threshold (ρ) are 10% and 70% respectively.

[0075] Table 2 Representative three-branch decision association rules (θ=10%, ρ=70%)

[0076]

[0077] Table 3 Non-representative three-way decision association rules (θ=10%, ρ=70%)

[0078]

[0079]

[0080] Based on the constructed OE3W concept lattice, representative three-branch decision association rules (Table 2) and non-representative three-branch decision association rules (Table 3) are extracted. For the OE3W concept (124, (ac, b)) in the OE3W concept lattice, the decision attribute set D is used to extract representative and non-representative three-branch decision association rules. The specific steps are as follows:

[0081] The extraction of representative three-branch decision association rules. First, the support threshold θ = 10% is used to extract OE3WL (T c ) to prune, and the pruned OE3WL (T c ) is recorded as OE3WL * (T c ). For OE3WL * (T c ) in any OE3W concept (124, (ac, b)), for attributes g and h in D, three-branch decision association rule The support and confidence of 66.67% both meet the set minimum support and confidence thresholds of 10% and 70%, so 66.67% are added to the set Ψ(T) storing representative three-branch decision association rules. * (T c ) and all the attributes in D, and finally all the representative three-branch decision association rules are obtained, as shown in Table 2.

[0082] Extraction of non-representative three-way decision association rules. For representative three-way decision association rules First find The corresponding connotation approximation set, yes The pseudo connotation of 50%, 66.67% are added to the set Φ(T) storing non-representative three-branch decision association rules. All representative three-branch decision association rules (Table 2) are traversed, and finally all non-representative three-branch decision association rules are obtained, as shown in Table 3.

[0083] The steps to obtain the intension approximation set of the intension (ac,b) of the OE3W concept (124,(ac,b)) are as follows: TEMP is the set of subsets of the attribute set (ac,b). For the attribute sets (c,b) and (a,b) in TEMP, there does not exist any parent concept (H,(O,S)) of (124,(ac,b)) such that Then add (c,b), (a,b), and (ac,b) into the connotation approximation set, where (c,b) and (a,b) are the pseudo-intensions of (ac,b).

[0084] 3. Rapid extraction of the three-cause rules of traffic accidents

[0085] Firstly, the attributes of the driver, vehicle, road and environment dimensions in traffic accidents are taken as conditional attributes, which are processed as conditional formal sub-contexts, and the severity classification of the accident is taken as the decision attribute; secondly, the obtained conditional formal sub-context of traffic accidents is gridded to obtain the three-branch concept lattice derived from the object, and the representative three-branch causal rules of traffic accidents are extracted using the OE3W concept lattice.

[0086] According to the method of extracting three-branch decision association rules based on OE3W concept lattice proposed in the present invention, it is applied to the mining of traffic accident causal rules. The traffic accident data of Bury Motorway in the UK from 2016 to 2020 are selected, and the three-branch causal rules of traffic accidents on Bury Motorway in the UK are extracted.

[0087] Table 4 Three-way decision association rules for Bury Motorway in the UK (partial)

[0088]

[0089]

[0090] Table 5 Two-way decision association rules for Bury Motorway in the UK (partial)

[0091]

[0092] Table 4 and Table 5 are the three-branch causal rules (part) and the corresponding two-branch causal rules (part) for traffic accidents on the Bury Motorway in the UK from 2016 to 2020. The relationship between the three-branch causal rules and the two-branch rules for traffic accidents was verified through experiments, which shows that the three-branch causal rules for traffic accidents are a supplement to the two-branch rules, can express knowledge that the two-branch rules cannot express, discover deeper connections and laws, and dig out a richer and more comprehensive traffic accident causal mechanism, which is helpful to deeply understand the causes of traffic accidents.

[0093] A fast extraction system for the three-cause rules of traffic accidents, such as Figure 7 As shown, the system includes the following modules:

[0094] The incremental algorithm constructs the OE3W concept lattice module 110, updates the three concepts derived from the object and the order relationship by analyzing the relationship between the three concept lattices derived from the objects before and after the newly added object g;

[0095] Extracting three-branch decision association rule module 120, firstly gives the definition of three-branch decision association rule, then divides the decision association rule into representative and non-representative three-branch decision association rule, and then gives the method of extracting representative and non-representative three-branch decision association rule based on OE3W concept lattice;

[0096] The module 130 for quickly extracting the three-cause rules of traffic accidents takes the attributes of the driver, vehicle, road and environment dimensions in the traffic accident as conditional attributes, processes them as conditional form sub-backgrounds, and classifies the severity of the accident as decision-making attributes; then the obtained traffic accident conditional form sub-backgrounds are gridded to obtain the three-branch concept grid derived from the object, and the OE3W concept grid is used to extract representative three-cause rules of traffic accidents.

[0097] The present invention adopts an incremental algorithm to construct the OE3W concept lattice, and extracts three-branch decision association rules on this basis. The method of extracting three-branch decision association rules based on the OE3W concept lattice is applied to the mining of traffic accident causal rules, which can effectively improve the efficiency of mining three-branch causal rules of traffic accidents. The proposed three-branch decision association rules based on the OE3W concept lattice are richer in semantics and more refined in expression than the two-branch decision association rules, and can express knowledge that the two-branch decision rules cannot express. Therefore, the method of extracting three-branch decision association rules based on the OE3W concept lattice is applied to the mining of traffic accident causal rules, which can mine a richer and more comprehensive traffic accident causal mechanism than the two-branch causal rules.

[0098] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for quickly extracting three-cause rules of traffic accidents, characterized in that: The steps include: S1. Use the incremental algorithm to construct the OE3W concept lattice. The incremental algorithm constructs the OE3W concept lattice by incrementally adding objects in the formal background to the formal background. c ) and OE3WL′(T c ) respectively represent the OE3W concept lattice corresponding to the formal background before and after the newly added object g; specifically, it includes the following steps: S11, add the OE3WL′(T c ) are divided into three categories: new OE3W concept, revised OE3W concept, and old OE3W concept; S12. Identify the necessary and sufficient conditions for the three types of OE3W concepts: new OE3W concepts, revised OE3W concepts, and old OE3W concepts; S13. Update the order relation. For each new OE3W concept (H′0, (O′0, S′0)), its standard generator is (H0, (O0, S0)), which must be the lower neighbor of the new OE3W concept (H′0, (O′0, S′0)). For each new OE3W concept (H′0, (O′0, S′0)), determining its neighbor set is divided into two steps: 1) Determine the candidate upper nearest neighbor set C = C1∪C2, where C1 and C2 are: C1={((O1,S1)) ＞ ′,(O1,S1))|(H1,(O1,S1))∈OE3WL(Tc),(H1,(O1,S1))>(H0,(O0,S0)),(H1,(O1,S1))is the modified OE3W concept}, C2=={(O′2,S′2) ＞ ′,(O′2,S′2))|(H2,(O2,S2))∈OE3WL(T c ),(H2,(O2,S2))>(H0,(O0,S0)),(O2,S2)∩(Y a ,Y b )≠(O2,S2)=(O′ 2, S′2)}; 2) The minimum OE3W concept set C′ in the candidate upper nearest neighbor set C is the upper nearest neighbor set of the new OE3W concept (H′0, (O′0, S′0)); For the modified OE3W concept in the upper neighbor set C′, the order relationship between the modified OE3W concept and its child concepts and parent concepts needs to be deleted. According to the above process, the order relationship is updated; S14. Through steps S11-S13, all OE3W concepts after the newly added object g can be updated, that is, all modified OE3W concepts, new OE3W concepts and the update of the order relationship after the added object g are found; continue to add the object g that has not been added to G, repeat steps S11-S14, traverse all objects in G, and finally obtain the conditional form sub-context T c OE3W concept grid OE3WL (T c ); S2, extracting three-branch decision association rules based on OE3W concept lattice; S3. Rapid extraction of the three-cause rules of traffic accidents, specifically including: first, taking the attributes of the driver, vehicle, road and environment dimensions in the traffic accident as conditional attributes, processing them as conditional form sub-backgrounds, and classifying the severity of the accident as decision attributes; secondly, building a grid with the obtained traffic accident conditional form sub-background to obtain the object-derived three-branch concept lattice, and using the OE3W concept lattice to extract representative three-branch cause rules of traffic accidents.

2. The method for rapidly extracting the three-cause rules of traffic accidents according to claim 1 is characterized by: In step S11, the specific steps include: S111、If (H′,(O′,S′))∈OE3WL′(T c ), and its intension (O′,S′) is not OE3WL(T c ), then (H′,(O′,S′)) is called a new OE3W concept; S112、If (H, (O, S))∈OE3WL(T c ), (H∪g,(O,S))∈OE3WL′(T c ), then (H, (O, S)) is called the modified OE3W concept; S113、If (H, (O, S))∈OE3WL(T c ), and (H,(O,S))∈OE3WL′(T c ), then (H,(O,S)) is called the old OE3W concept.

3. The method for rapidly extracting the three-cause rules of traffic accidents according to claim 1 is characterized by: In step S12, the following is specifically included: S121、If (H, (O, S))∈OE3WL(T c ), (H′,(O′,S′))∈OE3WL′(T c ), (O′, S′) = Then (H′, (O′, S′)) is called a new OE3W concept. In order to speed up the generation of new OE3W concepts, a standard generator is introduced. For any new OE3W concept (H′, (O′, S′)), is the standard generator of (H′,(O′,S′)); S122、If (H, (O, S))∈OE3WL(T c ), Right now Then it means that (H, (O, S)) is a modified OE3W concept; S123、If (H, (O, S))∈OE3WL(T c ), (H′,(O′,S′))∈OE3WL′(T c ), Then (H, (O, S)) is called the old OE3W concept.

4. The method for rapidly extracting the three-cause rules of traffic accidents according to claim 1 is characterized in that: In step S2, the specific steps include: S21. Definition of three-branch decision association rule: Let T = (G, A, I, D, J) be a decision form background. For any X, in if θ, ρ are thresholds set in advance, then are three-branch decision association rules, and their support and confidence are recorded as S22. Distinguish decision association rules into representative three-branch decision association rules and non-representative three-branch decision association rules: If are two three-way decision association rules of T, if and but Can be Z represents, denoted by For any If not present Then it is called is a representative three-branch decision association rule, otherwise it is called is a non-representative three-way decision association rule; S23. Extraction of representative three-branch decision association rules: First, use the support threshold θ to identify OE3WL(T c ) to prune, and the pruned OE3WL (T c ) is recorded as OE3WL * (T c ); for OE3WL * (T c ) in any OE3W concept (H, (O, S)), for any attribute Z in D, if it satisfies Then Add to the set of representative three-branch decision association rules Ψ(T); traverse each attribute in D and OE3WL * (T c ) for each OE3W concept, and finally obtain all the representative three-way decision association rules Ψ(T); S24. Extraction of non-representative three-branch decision association rules: For any representative three-branch decision association rule First, find the set of all intensional approximations of (O,S). (X,P) is called intensional approximation of (O,S) if and only if X * ∩P°=O * ∩S°=H, if (X,P) is a pseudo-intension of (O,S), then Then (X,P) is a pseudo-intension of (O,S), then Add the set of non-representative three-way decision association rules Φ(T); traverse the representative three-way decision association rules in Ψ(T), and finally obtain all non-representative three-way decision association rules Φ(T).

5. The method for rapidly extracting the three-cause rules of traffic accidents according to claim 4 is characterized by: In step S24, the specific steps for obtaining the intension approximation set of any OE3W concept (H, (O, S)) are as follows: TEMP is a set of subsets of the attribute set (O, S). For any attribute set (X, P) in TEMP, if there is no parent concept (H'0, (O'0, S'0)) of (H, (O, S)), such that Then add (X,P) to the connotation approximation set of the OE3W concept (H,(O,S)).

6. A system for quickly extracting the three-cause rule of traffic accidents according to any one of claims 1 to 5, characterized in that: The system includes the following modules: The incremental algorithm constructs the OE3W concept lattice module (110), updates the three concepts derived from the object and the order relationship by analyzing the relationship between the three concept lattices derived from the objects before and after the newly added object g; Extracting three-branch decision association rule module (120), firstly giving a definition of three-branch decision association rule, secondly distinguishing decision association rule into representative and non-representative three-branch decision association rule, and then giving a method for extracting representative and non-representative three-branch decision association rule based on OE3W concept lattice; The module (130) for quickly extracting the three-cause rules of traffic accidents takes the attributes of the driver, vehicle, road and environment dimensions in the traffic accident as conditional attributes, processes them as conditional form sub-backgrounds, and classifies the severity of the accident as decision attributes; then constructs a grid based on the obtained traffic accident conditional form sub-backgrounds to obtain the three-branch concept grid derived from the object, and uses the OE3W concept grid to extract representative three-cause rules of traffic accidents.

Images