KR20140134971A - Method of generating semantic models for FMEA documents - Google Patents

Abstract

A method of generating semantic models for failure mode and effect analysis (FMEA) documents comprises: a preprocessing step of extracting sentences from FMEA documents; a syntactic analysis step of generating a token where at least one part of speech is tagged by dividing the extracted sentences into tokens and analyzing parts of speech of the tokens; and a semantic analysis step of generating at least one semantic model by applying an extracted rule to the token where the part of speech is tagged.

Description

Field of the Invention [0001] The present invention relates to a method for generating a semantic model of an FMEA (Failure Mode and Effect Analysis (FMEA)

The present invention relates to an ontology, and more particularly, to a method for generating a semantic model of a Failure Mode and Effect Analysis (FMEA) document using an ontology.

When an engineer carries out a design, he or she often works with existing solutions to add new requirements. Effective management of product development knowledge in product development process is important to facilitate existing solutions.

An engineering document is unstructured, domain-specific for a specific area, contains sentences containing abbreviated, abbreviated, or alternate words, and therefore extracts or retrieves the necessary content from an engineering document. It is not easy to do.

Failure Mode and Effect Analysis (FMEA) is a type of engineering document used since the 1960s. FMEA refers to techniques that recognize potential failures and take precautions to minimize the cost of design changes. It is divided into System FMEA, Design FMEA, Process FMEA according to the application target, and is set as a standard in QS9000 and ISO9000, and it is recognized as a very important area in quality control by many companies. The output of the FMEA is managed as a tabular document with a template, and in the process of designing a new product, a similar failure mechanism is often applied by referring to existing FMEA documents.

The sentences in the FMEA document are short, diverse contents, various expressions of faults, abbreviations and omitted words are used. Therefore, it is difficult for the engineer to use the existing FMEA document to acquire the information, It takes a considerable amount of time. In many cases, search is not possible with existing keyword search, and accumulation of knowledge is also difficult.

An ontology is an optimal format for analyzing unstructured documents and structuring the analyzed information. In order to facilitate the understanding of the meaning of the contents of the FMEA document, it is necessary to generate the semantic model of the FMEA document using the ontology in order to process the FMEA document so that it can be easily retrieved and efficiently accumulate knowledge on the past history related to the failure.

It is an object of the present invention to provide a method for generating a semantic model of a Failure Mode and Effect Analysis (FMEA) document.

An object of the present invention is to provide a semantic model processing method of an FMEA document that generates a semantic model of an FMEA document, extends the generated semantic model, and integrates the extended semantic model.

It is an object of the present invention to provide an apparatus for processing a semantic model of an FMEA document.

In order to accomplish one object of the present invention, a method of generating a semantic model of a Failure Mode and Effect Analysis (FMEA) document comprises: a pre-processing step of extracting a sentence from an FMEA document; A syntactic analysis step of dividing the extracted sentence into tokens and analyzing the part of the token to generate at least one tag of the partly-tagged tokens; And a semantic analysis step of generating at least one semantic model by applying an extraction rule to the tagged token of the part-speech partner.

In one embodiment, the semantic analysis step includes: extracting at least one Concept instance included in the Semantic Model; And extracting a relation instance between at least one concept included in the semantic model.

In one embodiment, extracting the concept instance may comprise extracting a complex concept instance representing a plurality of adjacent concept instances.

In one embodiment, extracting the concept instance may further include extracting a failure concept instance by applying a failure concept instance extraction rule.

In one embodiment, the extraction rules may be stored in the domain knowledge store by the ontology 101 method.

In one embodiment, the form of the FMEA document may be XML.

In order to accomplish one object of the present invention, a semantic model processing method of a Failure Mode and Effect Analysis (FMEA) document is a method of generating at least one semantic model in units of cells of an FMEA document step; And generating an extended semantic model of the at least one semantic model.

In one embodiment, generating the at least one semantic model comprises: preprocessing a sentence in a cell of the FMEA document; A syntactic analysis step of dividing the extracted sentence into tokens and analyzing the parts of the token to generate at least one tagged token of the parts of speech; And a semantic analysis step of generating at least one semantic model by applying an extraction rule to the tagged token of the part-of-speech.

In one embodiment, the step of creating the extended semantic model may include the step of embodying a conceptual instance.

In one embodiment, generating the extended Semantic Model may further include extracting additional relationship instances using the causal relationship between the failure concept instances.

In order to accomplish one object of the present invention, a semantic model processing method of a Failure Mode and Effect Analysis (FMEA) document includes: generating at least one semantic model in units of cells of an FMEA document; And integrating the at least one semantic model.

In one embodiment, generating the at least one semantic model comprises: preprocessing a sentence in a cell of the FMEA document; A syntactic analysis step of dividing the extracted sentence into tokens and analyzing the parts of the token to generate at least one tagged token of the parts of speech; And a semantic analysis step of generating at least one semantic model by applying an extraction rule to the tagged token of the part-of-speech.

In one embodiment, the step of integrating the at least one semantic model comprises the steps of: (a) selecting two semantic models to be integrated if there are two semantic models to be integrated among the at least one semantic model; (b) terminating integration if there are no two semantic models to be integrated among the at least one semantic model; (c) selecting a similar concept instance among the two semantic models to be integrated as an integration target concept instance; (d) integrating the two semantic models to be integrated around the integration target concept instance; (e) re-executing steps (c) and (d) if all such similar concept instances are not integrated; (f) terminating integration if all such similar concept instances are consolidated and all of the at least one semantic model is attempted integration; And (g) re-performing steps (a) through (f) if all such similar concept instances are consolidated and all of the at least one semantic model is not attempted integration.

In one embodiment, the step (c) may include determining similarity between concept instances existing in the two semantic models to be integrated based on the similarity.

In one embodiment, the step (c) includes: selecting the first concept instance as an integration target concept instance if the similar concept instance is one first concept instance; And selecting a similar concept instance having a maximum number of connected concept instances among the similar concept instances as the integration target concept instance when there are a plurality of similar concept instances.

In one embodiment, step (d) includes: integrating the integration target concept instance; Integrating surrounding concept instances of the integration target concept instance; And extending the relationship of the integrated Semantic Model.

In one embodiment, a semantic model processing method of an FMEA document may be performed for all columns of a row of the FMEA document.

In one embodiment, a semantic model processing method of an FMEA document may be performed for all cells of the FMEA document.

In order to accomplish one object of the present invention, a semantic model processing method of a Failure Mode and Effect Analysis (FMEA) document includes: generating at least one semantic model in units of cells of an FMEA document; Generating an extended semantic model of the at least one semantic model; And integrating the at least one semantic model.

In order to accomplish one object of the present invention, a semantic model processing apparatus of a Failure Mode and Effect Analysis (FMEA) document includes a domain knowledge storage unit storing domain knowledge related to FMEA; A semantic model generating unit for generating at least one semantic model based on the information of the domain knowledge storage unit and the FMEA document; A semantic model extension unit for generating a semantic model having an expanded concept and relation of the at least one semantic model using information of the domain knowledge storage unit; And a semantic model integration unit for integrating the extended semantic model based on the degree of similarity using information of the domain knowledge storage unit.

If the semantic model of the Failure Mode and Effect Analysis (FMEA) document is generated using the present invention, knowledge about the past history related to the failure can be efficiently accumulated. Engineers can easily leverage knowledge of past history through the semantic model of the FMEA document, and product development time can be shortened.

FIG. 1 is a flowchart for processing a semantic model of a Failure Mode and Effect Analysis (FMEA) document according to an embodiment of the present invention.
2 is an example of an FMEA document.
3 is a block diagram illustrating a preprocessing process of an FMEA document according to an embodiment of the present invention.
4 is a block diagram illustrating a syntactic analysis step according to an embodiment of the present invention.
5 is a block diagram illustrating semantic analysis steps in accordance with an embodiment of the present invention.
Figure 6 is a block diagram illustrating the step of extracting a concept instance in accordance with an embodiment of the present invention.
FIG. 7 is a product list according to an information classification according to an embodiment of the present invention. FIG.
Figure 8 is a taxonomy according to an embodiment of the present invention.
Figure 9 is a block diagram illustrating the definition of a Concept and a Relation according to an embodiment of the present invention.
10 is a diagram illustrating a process of matching a token according to an embodiment of the present invention with a concept of a reference model according to an extraction rule.
11 is an extraction rule for a composite concept instance according to an embodiment of the present invention.
12 is an extraction rule of a failure concept instance according to an embodiment of the present invention.
13 is a semantic model of an FMEA document according to an embodiment of the present invention.
14 is a flowchart illustrating a step of generating an extended semantic model according to an embodiment of the present invention.
15 is a diagram illustrating a process of specifying a concept instance according to an embodiment of the present invention.
16 is an example of a process of specifying a concept instance according to an embodiment of the present invention.
Figure 17 is a semantic model of an FMEA document that includes a concrete conceptual instance according to an embodiment of the present invention.
18 is an extraction rule of an additional relation instance using a causal relationship between fault concept instances according to an embodiment of the present invention.
FIG. 19 is a block diagram showing a top-down relationship and a line-following relationship of a concept to which concept instances belong; FIG.
20 is a relation extraction rule according to a correlation of parts corresponding to a concept according to an embodiment of the present invention.
FIG. 21 is a semantic model of an FMEA document that includes a conceptual instance according to an embodiment of the present invention and whose relationship is extended.
22 is a flowchart illustrating a step of integrating a semantic model according to an embodiment of the present invention.
FIG. 23 is a flowchart illustrating a step of selecting a similar concept instance among the two semantic models to be integrated according to an embodiment of the present invention as an integration target concept instance.
Figure 24 is a flow diagram illustrating steps for integrating two semantic models to be consolidated around a similar concept instance according to an embodiment of the present invention.
25 is an illustration of an example 2500 of incorporating integration target concept instances of two semantic models to integrate in accordance with an embodiment of the present invention.
Figure 26 is an illustration of an example 2600 of incorporating conceptual instances around integration target concept instances of two semantic models to integrate in accordance with an embodiment of the present invention.
27 is a diagram illustrating a process of expanding the relationship of an integrated Semantic Model according to an embodiment of the present invention.
FIG. 28 is a diagram illustrating generation of a cell-based semantic model of an FMEA document and integration of the generated semantic model 2800 according to an embodiment of the present invention.
29 is an apparatus for processing a semantic model of an FMEA document according to an embodiment of the present invention.

For the embodiments of the invention disclosed herein, specific structural and functional descriptions are set forth for the purpose of describing an embodiment of the invention only, and it is to be understood that the embodiments of the invention may be practiced in various forms, The present invention should not be construed as limited to the embodiments described in Figs.

The present invention is capable of various modifications and various forms, and specific embodiments are illustrated in the drawings and described in detail in the text. It should be understood, however, that the invention is not intended to be limited to the particular forms disclosed, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. Similar reference numerals have been used for the components in describing each drawing.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between. Other expressions that describe the relationship between components, such as "between" and "between" or "neighboring to" and "directly adjacent to" should be interpreted as well.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having", etc., are intended to specify the presence of stated features, integers, steps, operations, components, parts, or combinations thereof, But do not preclude the presence or addition of other features, numbers, steps, operations, elements, parts or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted as either ideal or overly formal in the sense of the present application Do not.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The same reference numerals are used for the same constituent elements in the drawings and redundant explanations for the same constituent elements are omitted.

FIG. 1 is a flowchart for processing a semantic model of a Failure Mode and Effect Analysis (FMEA) document according to an embodiment of the present invention.

Referring to FIG. 1, in order to process a semantic model of an FMEA document, a sentence is first extracted from an FMEA document (step S110, pre-processing step). The step S110 will be described with reference to FIG. 2 and FIG.

Next, the extracted sentence is divided into tokens, and at least one part of the token is generated by analyzing the part of the token (step S120, syntactic analysis step). Step S120 will be described with reference to FIG.

Next, at least one semantic model is generated by applying the extraction rule to the tagged token of the part-of-speech (step S130, semantic analysis step). The step S130 will be described with reference to FIG.

Next, optionally, an extended semantic model of the at least one semantic model is generated (step S140). Step S140 will be described with reference to Fig.

Next, optionally, the at least one semantic model is integrated or the extended semantic model is integrated (step S150). Step S150 will be described with reference to FIG.

2 is an example of an FMEA document.

Referring to FIG. 2, the FMEA document 200 includes items such as a component function, a failure type, a failure effect, a failure cause, a countermeasure, and an action. Because these items have mutual causal relationships, there is a mutual causal relationship between the cells constituting the table of the FMEA document (200). It is possible to extend the semantic model generated based on cells through mutual causal relationship between cells.

3 is a block diagram illustrating a preprocessing process of an FMEA document according to an embodiment of the present invention.

Referring to FIG. 3, the FMEA document in the XML format includes a tag portion surrounded by " "and" > "for structuring meaningful information portion and meaningful information about a failure history. In the preprocessing step (S110) of the FMEA document, the XML document is parsed into the parts corresponding to each cell of the table of the FMEA document, and the XML tag is removed from the parsed sentence to generate a sentence having the meaning of the past fault history (300) of the FMEA document to be extracted. For example, the tag is removed from the FMEA document 310 and the sentence "Rotate FAN", "FAN drive motor not working", "Air flow FAN disabled" Extracted "sentences 320 of" MOTOR does not operate due to lack of torque when high load is applied to FAN ".

4 is a block diagram illustrating a syntactic analysis step according to an embodiment of the present invention.

Referring to FIG. 4, the extracted sentences 320 are divided into tokens constituting a sentence, and the part of speech of the divided token is analyzed to generate a token 410 having at least one part of speech. For example, the extracted sentences 320 may be sent as "FAN", "W", "ROTATE", "SHIFT", "K", "FAN", "DRIVE", "MOTOR" Quot ;, "back ", and" The results of tagging parts of speech are "FAN / ncn", "/ jco", "rotation / ncpa", "shiki / xsva", "k / etn", "FAN / ncn" / ncn "," act / ncpa "," not / mag "," re / pvg "," ㅁ / etn " ncn is a normal noun, ncpa is a synonym noun, xsva is a verb-derived suffix, etn is a noun-typed suffix, mag is a general adverb, pvg is a general verb, and jco is a subject exam.

5 is a block diagram illustrating semantic analysis steps in accordance with an embodiment of the present invention.

Referring to FIG. 5, in a semantic analysis step (step S130), at least one concept instance included in the semantic model is extracted (step S131). The step S131 will be described with reference to FIG.

Next, a relation instance between at least one concept included in the semantic model is extracted (step S132). The relationships are inherited from at least one concept instance extracted in step S131, that is, a relation instance is extracted to generate a basic semantic model.

Figure 6 is a block diagram illustrating the step of extracting a concept instance in accordance with an embodiment of the present invention.

Referring to FIG. 6, in order to extract a concept instance (step S131), the partly tagged token 410 is matched with the concept of the reference model according to an extraction rule (step S133). The step S133 will be described with reference to FIG.

Next, an instance of the matched concept is created (step S134). When a token is matched with one concept on the reference model, an instance of a matched concept (hereinafter referred to as a concept instance) is generated and used as a node constituting the semantic model. Since the token 410 with tagged parts of speech includes at least one token, there can be at least one concept instance corresponding to at least one token.

Next, a complex concept instance representing a plurality of adjacent concept instances is extracted (step S135). The step S135 will be described with reference to FIG.

Next, the failure concept instance is extracted by applying the failure concept instance extraction rule (step S136). The step S136 will be described with reference to FIG.

FIG. 7 is a product list according to an information classification according to an embodiment of the present invention. FIG.

Referring to FIG. 7, the article 700 according to the information classification analyzes typical parts of speech according to the information used in the FMEA. The information was classified into 10 kinds (device, function, failure, attribute, facility, processor, external environment, material, unit, feature). The article classification according to the information classification (700) analyzed the parts of the words representing each information.

Figure 8 is a taxonomy according to an embodiment of the present invention.

Referring to FIG. 8, a classification system 800 that takes into account the range of information handled by the FMEA based on article 700 based on information classification. The classification system 800 may be implemented on a reference model. The reference model of the classification system 800 includes a device, a function, a failure, a facility, an environmental object, a unit, a property, a process, , Material (Material), Recommended Action, and Taken Action.

Figure 9 is a block diagram illustrating the definition of a Concept and a Relation according to an embodiment of the present invention.

Referring to FIG. 9, the definition of concept and relationship 900 includes eleven sub-concepts of the reference model on classification system 800 and includes a relationship between the eleven sub-concepts of the reference model. For example, the relationship centering on the concept of "device" is as follows. There is a "has_function" relationship between the concepts of "device" and "function". There is a "has_part" relationship between the "device" concept and the "device" concept itself. There is a "has_material" relationship between the "device" concept and the "material" concept. There is a "has_process" relationship between the "device" and "process" concepts. There is a "has_property" relationship between the "device" concept and the "attribute" concept. There is a "has_failure" relationship between the "device" concept and the "failure" concept. There is a "has_object" relationship between the "functional" concept and the "device" concept. Since the remaining structure of FIG. 9 can be understood in view of the above description, the description is omitted.

10 is a diagram illustrating a process of matching a token according to an embodiment of the present invention with a concept of a reference model according to an extraction rule.

Referring to FIG. 10, the domain knowledge storage unit in which the domain knowledge is stored by the domain expert includes a database related to the conceptual synonyms / synonyms using concepts such as a reference model for concepts and a component specification (BOM), a past FMEA, .

The parts of the tagged tokens 1010 ("FAN / ncn", "drive / ncpa", "MOTOR / ncn", "act / ncpa", " ) And performs matching with the concept or concept of the classification system 1020 or similar concept on a token basis. The " FAN "token of the partly tagged tokens 1010 is a sub-concept of the " assembly" concept 1022, a sub- Quot; Fan "concept, which is an agreement or similar concept 1024 of " The "driving" token of the tokens 1010 with the tagged parts of speech is matched with the "driving" concept, which is the agreement or similarity concept 1028 of the "function" concept 1027 of the classification system 1020. The "MOTOR" token of the tagged tokens 1010 is a "motor" concept 1025 that is a sub-concept of the &Quot; engine ", which is an agreement or similar concept 1026 of " motions ". The "action" token of the tokens 1010 with the tagged parts of speech is matched with the " action "concept of the consent or similar concept 1028 of the concept 1027 of the classification system 1020. Since the remaining structure of FIG. 10 can be understood in view of the above description, the description is omitted.

11 is an extraction rule for a composite concept instance according to an embodiment of the present invention.

Referring to FIG. 11, if the concept instance is matched with the concept of a " device "after step S134, a DCI (Device Concept Instance) Concept Instance), FTCI (Feature Concept Instance) if it matches with the concept of "Feature", and Failure Concept Instance (FLCI) if it matches with the concept of "Failure".

The instances of all concepts are not extracted only by the matched concept instances generated by the execution up to step S134. An extraction rule 1100 of a complex concept instance is used to extract a complex concept instance (a step S135) in which a plurality of instances of concepts are written adjacent to each other and recognized as a concept.

In an example of extraction rule 1100 of a complex concept instance, when DCI1 and DCI2 are adjacent to each other, a concept instance of DCI1 and DCI2 extracts the concept of DCI2. In the actual example, if you apply the above rule, DCI1 will be set to "<Fan> FAN </ Fan>" and DCI2 will be set to "<Motor>" if the input is "<Fan> FAN </ Fan> <Motor> MOTOR </ motor>, so that the compound concept instance for the entire input has the concept of "motor" which is the concept of DCI2. The remaining structure of FIG. 11 can be understood in view of the above description, and a description thereof will be omitted.

12 is an extraction rule of a failure concept instance according to an embodiment of the present invention.

Referring to FIG. 12, a concept instance generated by the execution up to step S135 does not include a failure concept instance (FLCI). FLCI extraction is performed using the extraction rule 1200 of the failure concept instance for analyzing various failure expression styles of the FMEA (step S136).

13 is a semantic model of an FMEA document according to an embodiment of the present invention.

13, the semantic model 1300 generated by the execution up to step S130 includes a concept instance 1310 of the concept of "fan", a concept "instance of driving" 1320 of the concept of "function" Concept " concept "MOTOR " concept 1330 and a" failure "concept FLCI 1340 are extracted. The concept instance and relation definition (800) is applied to these concept instances to extract relationship instances.

Referring to FIGS. 13 and 8, the object of the "has_object" relationship of the "driving" concept instance 1320 of the "function" concept is extracted to the "FAN" concept instance 1310 of the "fan" concept. The object of the "related_with" relation of the " failure "concept of the" failure "FLCI 1340 is extracted into the" The object of the "has_function" relation of the "MOTOR" concept instance 1330 of the "motor" concept is extracted into the "driving" concept instance 1320 of the "function" concept.

14 is a flowchart illustrating a step of generating an extended semantic model according to an embodiment of the present invention.

Referring to FIG. 14, in order to generate an extended semantic model (step S140), a concept instance is first specified (step S141). The step S141 will be described with reference to FIG. 15 to FIG.

Next, an additional relationship instance is extracted using the causal relationship between the failure concept instances (step S142). The step S142 will be described with reference to Figs. 18 to 21. Fig.

15 is a diagram illustrating a process of specifying a concept instance according to an embodiment of the present invention.

Referring to FIG. 15, the refinement process 1500 of a concept instance includes a concept instance 1511, 1512 of a semantic model 1510 of an instance domain, a hierarchical level 1512 of a concept belonging to a classification system 1520 of a concept domain, Quot; R3 "relationship 1513 between the concept instances 1511 and 1512 is refreshed.

The concept domain classification system 1520 includes a concept "C1" 1523 and a concept "C2" 1524 as a sub concept of the "Root" concept 1527. The classification system 1520 of the concept domain includes the concepts of "C11" and "C12" as subcategories of the concept "C1" The classification system 1520 of the concept domain includes the concepts of "C21" and "C22" as sub concepts of the concept "C2" The classification system 1520 of the concept domain includes the concepts of "C111" and "C112" The classification system 1520 of the concept domain includes the concept "C211 " 1522 as a sub concept of the concept" C21 ". The classification domain 1520 of the concept domain has a relationship between the "R1" relationship 1525 between the "C1" concept 1523 and the "C2" concept 1524 and between the "C112" concept 1521 and the "C211" R2 "relation 1526. &lt; / RTI &gt;

The structural level of the classification domain of the concept domain (1520) is a measure of the structural distance from the concept of "root". The concept of "root" has a structural level value of zero. The concept "C1" concept (1523) and concept "C2" (1524) have a structural level value of 1. The concepts "C11", "C12", "C21" and "C22" have a structural level value of 2. The concepts of "C111 "," C112 ", and "C211"

The semantic model 1510 of the instance domain includes an " I1 "concept instance 1511 and an" I2 "concept instance 1512, "Relation instance 1513. &lt; / RTI &gt;

Quot; I1 "concept instance 1511 belongs to" C112 "concept 1521 (1531), and the" I2 & Belongs to the " C2 "concept 1524 (1532), and the" R3 "relation instance 1513 belongs to the" R1 &

The refinement process 1500 of the concept instance searches the classification domain 1520 of the concept domain to find a class 1520 having a structural level higher than 1, which is the structural level value of the " C2 "concept 1524 to which the &Quot; C211 "concept 1522 having a structure level value of 3 with the " C112 " concept 1521 and the" R2 "relation 1526 to which the" I1 "concept instance 1511 belongs, The concept of the concept instance 1512 is incorporated into the concept " C211 " 1522 and is refreshed so that the "R3" relation instance 1513 belongs to the "R2" relationship 1526.

16 is an example of a process of specifying a concept instance according to an embodiment of the present invention.

Referring to FIG. 16, an example 1600 of the conceptualization of a conceptual instance includes a classification system 1620 of a concept domain and a semantic model 1610 of an instance domain.

The concept domain classification system 1620 includes the concepts of "device" concept 1623 and "function" 1624 as sub-concepts of the "root" concept 1627. The classification domain 1620 of the concept domain includes the concept "fan" and the concept "motor" 1621 as a subset of the "device" The classification system 1620 of the concept domain includes the concept "rotate pan" 1622 as a sub-concept of the "function" concept. The classification domain 1620 of the concept domain includes a "has_function" relationship 1625 between the "device" concept 1623 and the "function" concept 1624 and a concept " Has a "has-function"

The instance domain semantic model 1610 includes a "Motor" concept instance 1611 and a "drive" concept instance 1612. The instance domain semantic model 1610 does not include a relationship.

In the case of the semantic model 1610 of the instance domain generated after the execution of step S130, the "Motor" concept instance 1611 belongs to the "motor" concept 1621 (1631) Belongs to the concept "function" 1624 (1632).

Search for a classification domain 1620 of a concept domain yields a "Motor" concept instance 1620 of a sub-concept having a structural level value higher than 1, which is the structural level of the " The concept of the "drive" concept instance 1612 is referred to as "rotating the fan " since there is a" rotate fan "concept 1622 with a" motor " Quot; concept 1622 (1633).

Since the semantic model 1610 of the instance domain does not include a relationship instance, no relationship change is made.

Figure 17 is a semantic model of an FMEA document that includes a concrete conceptual instance according to an embodiment of the present invention.

Referring to FIG. 17, there is shown a semantic model 1700 of an FMEA document that contains a materialized concept instance that is the result of an example 1600 of the instantiation of a concept instance.

The semantic model 1700 of the FMEA document including the instance of the concrete concept created by performing the operations up to step S141 is replaced by the semantic model 1700 of the concept of "function" Includes a "drive" concept instance 1720 of the "spin pan" The semantic model 1700 of the FMEA document that contains an instance of the materialized concept except for the " drive "concept instance 1720 of the" spin pan " Is the same as the semantic model 1300.

18 is an extraction rule of an additional relation instance using a causal relationship between fault concept instances according to an embodiment of the present invention. In order to extract additional relationship instances using the causal relationship between instances of failure concepts (step S142), additional rules for extracting relation instances 1800 using causal relationships between the failure concept instances can be used.

Referring to FIG. 18, when there are a plurality of failure concept instances (FLCIs), a plurality of failure concept instances (FLCIs) Additional relationship instances are extracted according to the extraction rule 1800 of the additional relationship instance that analyzed the expression representing the causal relationship between the failure concept instances.

For example, if the fault concept instance is linked with "/ nbn" because of "FLCI1 due to / nbn FLCI2", the "has_effect (FLCI1, FLCI2)" relationship between FLCI1 and FLCI2 is extracted.

FIG. 19 is a block diagram showing a top-down relationship and a line-following relationship of a concept to which concept instances belong; FIG. In order to extract additional relationship instances using the causal relationship between instances of failure concepts (step S142), it is possible to use the up-and-down relationship and the pre-relationship (1900) of the concept to which the concept instances on the classification system belong.

Referring to FIG. 19, the up-down and up-down relationships 1900 of the concept to which the concept instances on the classification system belong include the concept domain classification system 1910 and the instance domain semantic model 1920.

The concept "B" 1911 in the concept domain classification system 1910 includes the first concept 1912 and the concept "A" 1913 as sub-concepts. The second concept (1914) in the concept domain classification system (1910) includes the concept "C" (1915) and the concept "D" (1916) as subcategories. The classification domain 1910 of the concept domain includes the relationship 1917 between the concept "C" 1915 and the concept "1916". The concepts 1911 to 1916 refer to the sub-concept of the device concept.

The semantic model 1920 of the instance domain includes the FLCI 1921 of the "A" concept instance, the FLCI 1922 of the "B" concept instance, the FLCI 1923 of the "C" 1924).

 The analysis of the classification domain 1910 of the concept domain reveals that the parts corresponding to the concept A 1913 are broken down into parts corresponding to the concept B 1911 which is a superordinate concept of the concept A 1913, Can affect the function of the system. Using this, a "has_effect" relationship 1925 between the FLCI 1921 of the instance of the concept "A" and the FLCI 1922 of the instance of the "B" concept can be extracted.

Analysis of the classification domain 1910 of the concept domain reveals that the component corresponding to the "C" concept 1915 has a component " D "concept 1916, 1915) may affect the function of the trailing component corresponding to the "D" concept 1916. [ Using this, a "has_effect" relationship 1926 between the FLCI 1923 of the instance of the "C" concept and the FLCI 1924 of the instance of the "D" concept can be extracted.

20 is a relation extraction rule according to a correlation of parts corresponding to a concept according to an embodiment of the present invention.

Referring to FIG. 20, there is a relation extraction rule (2000) according to the correlation of parts corresponding to concept, for applying the up-down relationship and the line-following relation (1900) of concepts on the classification system.

In the case of an association of faults indicating the failure of part B corresponding to the trailing concept due to the failure of part A corresponding to the preceding concept, the relation extraction rule 2010 is as follows. (Has_failure (DCI1, FLCI1)) as a failure concept instance of DCI1 with DCI1 having the function of FCI1 (has_function (DCI1, FCI1)) and DCI2 as the object of FCI1 (has_object (FCI1, DCI2) (Has_failure (DCI2, FLCI2)) and "has_effect (FLCI1, FLCI2)" can be extracted when the FLCI2 exists in the failure concept instance of DCI2. If the DCI1 fails (FLCI1), the function of FCI1 can not be performed, and the DCI2 of the FCI1 also can not function. Therefore, the relation of "has_effect (FLCI1, FLCI2)" can be extracted because it affects FLCI2.

An example 2020 of the relationship extraction rule 2010 is as follows. If the sentence is "FAN driven motor not working", the semantic model relationships created based on the sentence are "has_function (MOTOR, drive)", "has_object (drive, FAN)", "has_failure ) "And " has_failure (FAN, FL2) ". Based on this, we can extract the "has_effect" relationship. The remaining case of FIG. 20 can be understood in light of the above description, and a description thereof will be omitted.

FIG. 21 is a semantic model of an FMEA document that includes a conceptual instance according to an embodiment of the present invention and whose relationship is extended.

Referring to FIG. 21, the Semantic Model 2100 of the FMEA document including the conceptual instance that has been embodied and the relationship expanded includes a Semantic Model 1700 of the FMEA document including an instance of the materialized concept. Since there is no FLCI that is the subject of the "has_failure" relationship of the "FAN" concept instance 2113 of the "fan" concept, the rules are applied after generating the fault 1 FLCI 2111. The relation extracted by the relation extraction rule 2010 is a relation 2121 between "has_failure (FAN, fault 1)" and "relation_with (not working, fault 1)" relationship 2112.

22 is a flowchart illustrating a step of integrating a semantic model according to an embodiment of the present invention.

Referring to FIG. 22, in order to integrate at least one semantic model (step S150), it is determined whether there is a semantic model to be integrated into at least one semantic model (step S151). If there is only one semantic model, the integration of the at least one semantic model ends because integration of the semantic model is not necessary.

Next, when there is a semantic model to be integrated into at least one semantic model, two semantic models to be integrated are selected (steps S151 and S152).

Next, a similar concept instance among the two semantic models to be integrated is selected as an integration target concept instance (step S153). The step S153 will be described with reference to Figs. 23 and 25. Fig.

Next, two semantic models to be integrated around the integration target concept instance are integrated (step S154). The step S154 will be described with reference to FIG. 24, FIG. 26, and FIG.

Next, it is determined whether all similar concept instances are integrated (step S155). If all similar concept instances are not integrated, steps S153 and S154 are executed again.

Next, if all similar concept instances are integrated, it is determined whether all semantic models of the at least one semantic model have been integrated (step S156). The steps S151 to S155 are repeated until all the semantic models of the at least one semantic model are attempted to be integrated. If all of the semantic models are attempted to be integrated, the integration of the at least one semantic model is terminated.

FIG. 23 is a flowchart illustrating a step of selecting a similar concept instance among the two semantic models to be integrated according to an embodiment of the present invention as an integration target concept instance.

Referring to FIG. 23, in order to select a similar concept instance among the two semantic models to be integrated as an integration target concept instance (step S153), a concept instance having the same concept as the lower concept of the device concept is searched (step S161). The concept instance to be integrated is limited to instances of sub-concepts of device concepts.

Next, it is determined whether there is a concept instance having the same concept that is searched (step S162). If there is no concept instance having the same concept that is searched for, the step of selecting the similar concept instance among the two semantic models to be integrated (step S153) is ended.

Next, if there is a concept instance having the same concept as found in step S162, it is determined whether the concept instances having the same concept are similar based on the degree of similarity (step S163).

Next, the number of similar concept instances is determined (step S164). If there is no similar concept instance, the step of selecting a similar concept instance among the two semantic models to be integrated as an integration target concept instance (step S153) is ended.

Next, if the similar concept instance is one first concept instance, the first concept instance is selected as the integration target concept instance (step S165).

Next, when there are a plurality of similar concept instances, a similar concept instance having a maximum number of connected concept instances among similar concept instances is selected as an integration target concept instance (step S166).

Figure 24 is a flow diagram illustrating steps for integrating two semantic models to be consolidated around a similar concept instance according to an embodiment of the present invention.

Referring to FIG. 24, in order to integrate two semantic models to be integrated around a similar concept instance (step S154), the integration target concept instance is first integrated (step S171).

Next, concept instances around the integration target concept instance are integrated (step S172).

Next, the relationship of the integrated Semantic Model is extended (step S173).

25 is an illustration of an example 2500 of incorporating integration target concept instances of two semantic models to integrate in accordance with an embodiment of the present invention.

25, the " FAN "concept of the" fan "of the first semantic model 2510 is used to integrate the first semantic model 1 2510 and the second semantic model 2 2520, The concept instance 2511 of the concept of "fan" of the second semantic model 2520 and the concept instance 2521 of the second semantic model 2520 are searched (step S161) with a concept instance having the same concept as the sub-concept of the "device" concept.

The determination of the similarity of the concept instances having the same concept on the basis of the similarity degree (step S163) uses Equations (1) and (2).

Wm in Equation (1) denotes a weight of a concept instance. If one concept instance (I _i ) and another concept instance (I _j ) are composed of one word and Wm = 1 when the one word is matched, one concept instance (I _i ) I _j ) does not include a matched word Wm = 0, and one concept instance (I _i ) and another concept instance (I _j ) contain a plurality of matched words, the rightmost matched word Wm = 0.55, and Wm = 0.45 for the remaining matched words.

When the TScore _ij value in Equation (1) is 0.5 or more, the IScore _i value in Equation (2) is 0.5 or more. In this case, one concept instance (Ii) and another concept instance (Ij) are defined as similar concept instances. Conversely, when the TScore _ij value in Equation (1) is smaller than 0.5, the IScore _i value in Equation (2) becomes smaller than 0.5. In this case, it is defined that one concept instance (Ii) and another concept instance (Ij) are not similar concept instances.

Since the concept instance is the same as the " FAN " concept instance 2521 of the "fan" concept and the FAN concept instance 2521 of the concept of "fan ", the TScore value is 1 and the IScore value is also 1 , The "FAN" concept instance 2511 of the "fan" concept is determined to be a concept instance similar to the "FAN" concept instance 2521 of the "fan" concept (step S163).

FAN "concept instance 2511 of the" fan "concept is selected as the integration target concept instance (step S165) because there is one" FAN &

FAN "concept instance 2511 of the first semantic model 2510 and the" FAN "concept instance 2521 of the" fan "concept of the second semantic model 2520, Concept " FAN "concept instance 2531 is generated.

Figure 26 is an illustration of an example 2600 of incorporating conceptual instances around integration target concept instances of two semantic models to integrate in accordance with an embodiment of the present invention.

Referring to FIG. 26, the degree of similarity between instances having the same approximate distance in descending order of the proximity distance (D) from the "FAN" concept instance 2531 of the concept of "pan" do. The approximate distance is expressed by the number of edges from the instance (I), and the sign has a positive sign if it has an incoming relationship from I and a negative sign if it has an outgoing relationship to I.

In an example 2600 of incorporating concept instances around integration target concept instances of the two semantic models to integrate, the " FAN "concept instance 2611 of the" Concept "instance" 2511 of the concept of "import air" having an incoming relationship from the concept instance 2611 and the "inoperative" concept instance 2612 of the concept of " Value. The "air" concept instance 2622 of the "external environment" concept has a value of D = + 2 because it has an incoming relationship from the "flow" concept instance 2511 of the concept of " Concept " concept "concept instance 2521 has a value of D = + 2 because it has an incoming relationship from the " inoperative" concept instance 2612 of the " Since the remaining structure of FIG. 26 can be understood in view of the above description, description thereof will be omitted.

In the example of FIG. 26 (2600), there is no instance to join because the similarity between instances having the same approximate distance D is calculated.

27 is a diagram illustrating a process of expanding the relationship of an integrated Semantic Model according to an embodiment of the present invention.

Referring to FIG. 27, a process 2700 of expanding the relationship of the integrated Semantic Model can be performed by applying the relation extraction rule 2010 of the relation extraction rule 2000 according to the correlation of the parts corresponding to the concept. Has_effect "relationship 2713 for the" inoperative "concept instance 2711 of the" faulty fan "concept from the" inactive "concept instance 2712 of the" motor "concept.

FIG. 28 is a diagram illustrating generation of a cell-based semantic model of an FMEA document and integration of the generated semantic model 2800 according to an embodiment of the present invention.

Referring to Fig. 28, the FMEA document 2801 is composed of at least one cell constituting a table. First, semantic models 2821, 2822, 2823, 2824, 2841, 2842, 2843 and 2844 are generated based on the cells 2811, 2812, 2813, 2814, 2831, 2832, 2833 and 2834.

The first semantic model set 2820 including the semantic models 2821, 2822, 2823, and 2824 for the same row among the generated semantic models 2821, 2822, 2823, 2824, 2841, 2842, 2843, The semantic model 2851 of the integrated FMEA row is generated by using the integration method described in FIGS.

An mth semantic model set 2840 including semantic models 2841, 2842, 2843 and 2844 for the same row among the generated semantic models 2821, 2822, 2823, 2824, 2841, 2842, 2843, The semantic model 2852 of the merged FMEA m row is generated using the integration method described in Figs.

The semantic model set 2850 of each row of the FMEA including the generated FMEA 1 row semantic model 2851 and FMEA m row semantic model 2852 is integrated using the integration method described in FIGS. Thereby generating a semantic model 2860 of the entire FMEA.

29 is an apparatus for processing a semantic model of an FMEA document according to an embodiment of the present invention.

29, the apparatus for processing a semantic model of an FMEA document includes a domain knowledge storage unit 2914 and a semantic model generation unit 2911. [ The semantic model processing apparatus of the FMEA document optionally includes a semantic model extension unit 2912. The semantic model processing apparatus of the FMEA document optionally includes a semantic model integrating unit 2913.

The domain knowledge storage unit 2914 includes a first step of determining a domain and a scope of an ontology to be developed, a second step of selectively reusing the existing ontology, a third step of listing important words of the ontology to be developed, And the fourth step of defining a class and a hierarchy of the ontology to be developed by the ontology development 101 method that is widely known to a general technician 2922 are stored in advance and the domain knowledge 2922 is transmitted to the semantic model generating unit 2911, the semantic model extending unit 2912 and the semantic model integrating unit 2913. The domain knowledge 2922 of the domain knowledge storage unit 2914 may be generated by an ordinary descriptor or an expert of a domain of the ontology to be developed.

The semantic model generation unit 2911 extracts a sentence from the FMEA document 2921, divides the extracted sentence into tokens, analyzes the part of the token, generates at least one part of speech tagged token, Apply extraction rules to tokens to create at least one semantic model.

The semantic model extension unit 2912 specifies the concept instance of the at least one semantic model, extracts additional relationship instances using the causal relationship between the failure concept instances, and extracts the extended semantic model of the at least one semantic model .

The semantic model integrating unit 2913 generates an integrated semantic model 2923 that integrates the at least one semantic model or the extended semantic model.

The domain knowledge storage unit 2914, the semantic model generation unit 2911, the semantic model extension unit 2912, and the semantic model integration unit 2913 may be implemented through software, may be implemented through hardware, And hardware.

The semantic model of the FMEA document enables engineers to shorten their product development time, thereby enhancing product competitiveness.

Although the preferred embodiments of the present invention have been disclosed for illustrative purposes, those skilled in the art will appreciate that various modifications, additions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims. It will be understood that the invention may be modified and varied without departing from the scope of the invention.

Claims (20)

Failure Mode and Effect Analysis (FMEA) A preprocessing step for extracting sentences from a document;
A syntactic analysis step of dividing the extracted sentence into tokens and analyzing parts of the token to generate at least one tag of the partly-tagged tokens; And
And a semantic analysis step of generating at least one semantic model by applying an extraction rule to the tagged token of the part-of-speech to generate a semantic model of the FMEA document. The method according to claim 1,
Wherein the semantic analysis step comprises:
Extracting at least one Concept instance included in the Semantic Model; And
And extracting a relation instance between at least one concept included in the semantic model. 3. The method of claim 2,
Wherein extracting the concept instance comprises extracting a complex concept instance representing a plurality of adjacent concept instances. &Lt; RTI ID = 0.0 &gt; 8. &lt; / RTI &gt; The method of claim 3,
Wherein the step of extracting the concept instance further includes extracting a fault concept instance by applying a fault concept instance extraction rule to the FMEA document. The method according to claim 1,
Wherein the extraction rule is stored in a domain knowledge storage unit by an ontology 101 method. The method according to claim 1,
Wherein the form of the FMEA document is XML. Generating at least one semantic model in units of cells of a Failure Mode and Effect Analysis (FMEA) document; And
And generating an extended Semantic Model of the at least one Semantic Model. 8. The method of claim 7,
Wherein the generating the at least one semantic model comprises:
A preprocessing step of extracting a sentence from a cell of the FMEA document;
A syntactic analysis step of dividing the extracted sentence into tokens and analyzing the parts of the token to generate at least one tagged token of the parts of speech; And
And a semantic analysis step of generating at least one semantic model by applying an extraction rule to the tagged token of the part-of-speech. 9. The method of claim 8,
Wherein the step of generating the extended Semantic Model includes a step of specifying a concept instance. 10. The method of claim 9,
Wherein the step of generating the extended Semantic Model further comprises extracting additional relationship instances using a causal relationship between the fault concept instances. Generating at least one semantic model in units of cells of a Failure Mode and Effect Analysis (FMEA) document; And
And integrating the at least one Semantic Model. &Lt; Desc / Clms Page number 22 &gt; 12. The method of claim 11,
Wherein the generating the at least one semantic model comprises:
A preprocessing step of extracting a sentence from a cell of the FMEA document;
A syntactic analysis step of dividing the extracted sentence into tokens and analyzing the parts of the token to generate at least one tagged token of the parts of speech; And
And a semantic analysis step of generating at least one semantic model by applying an extraction rule to the tagged token of the part-of-speech. 12. The method of claim 11,
Wherein integrating the at least one semantic model comprises:
(a) selecting two semantic models to be integrated when there are two semantic models to be integrated among the at least one semantic model;
(b) terminating integration if there are no two semantic models to be integrated among the at least one semantic model;
(c) selecting a similar concept instance among the two semantic models to be integrated as an integration target concept instance;
(d) integrating the two semantic models to be integrated around the integration target concept instance;
(e) re-executing steps (c) and (d) if all such similar concept instances are not integrated;
(f) terminating integration if all such similar concept instances are consolidated and all of the at least one semantic model is attempted integration; And
(g) re-executing the steps (a) through (f) if all the similar concept instances are integrated and all of the at least one semantic model is not attempted integration. Model processing method. 14. The method of claim 13,
The step (c)
And determining whether there is a similarity between concept instances existing in the two semantic models to be integrated based on the degree of similarity. 15. The method of claim 14,
The step (c)
Selecting the first concept instance as an integration target concept instance if the similar concept instance is one first concept instance; And
And selecting a similar concept instance having a maximum number of connected concept instances among the similar concept instances as an integration target concept instance when the similar concept instance has a plurality of similar concept instances. 14. The method of claim 13,
The step (d)
Integrating the integration target concept instance;
Integrating concept instances around the integration target concept instance; And
And expanding the relationship of the integrated Semantic Model to the Semantic Model. 12. The method of claim 11,
Wherein the steps are performed for all columns of one row of the FMEA document. 12. The method of claim 11,
Wherein the steps are performed for all cells of the FMEA document. Generating at least one semantic model in units of cells of a Failure Mode and Effect Analysis (FMEA) document;
Generating an extended semantic model of the at least one semantic model; And
And integrating the extended Semantic Model. &Lt; Desc / Clms Page number 21 &gt; A domain knowledge storage for storing domain knowledge related to Failure Mode and Effect Analysis (FMEA);
A semantic model generating unit for generating at least one semantic model based on the information of the domain knowledge storage unit and the FMEA document;
Alternatively, the semantic model extension unit may generate a semantic model that extends the concept and relationship of the at least one semantic model using information of the domain knowledge storage unit. And
Optionally, the semantic model integration unit integrates the extended semantic model based on the degree of similarity using information of the domain knowledge storage unit.

Images