CN104182454A - Multi-source heterogeneous data semantic integration model constructed based on domain ontology and method - Google Patents

Abstract

The invention discloses a multi-source heterogeneous data semantic integrated model constructed based on domain ontology. The multi-source heterogeneous data semantic integration model comprises a local ontology constructing module, a domain ontology merging module and a semantic inquiring dynamic propagation and protocol module. A multi-source heterogeneous data semantic integrating method comprises the following steps: constructing the domain ontology through an ontology merging technique and establishing a semantic mapping relation between a data source and the local ontology as well as between the local ontology and the domain ontology; combining the complementary advantage of social label and ontology on knowledge representation and performing inquiring protocol and propagation for the semantic inquiring request of the user, thereby generating a formal semantic inquiring statement; respectively inquiring a plurality of data sources, performing duplicate removal and aggregation optimization on the inquiring result, and lastly feeding back to the user. The invention provides an aggregation heterogeneous data semantic integration model constructed based on domain ontology and the method through the construction and mapping of the domain ontology, semantic inquiring propagation and result aggregation optimization.

Description

The integrated models and methods of multi-source heterogeneous data semantic building based on domain body

Technical field

The crossing domain that the invention belongs to the semantic integrated and oil-gas exploration Knowledge Discovery of isomeric data, relates in particular to a kind of multi-source heterogeneous data semantic integrated model and method building based on domain body.

Background technology

Along with industrialization and information-based deepening continuously of merging, oil-gas exploration enterprise in long-term production run, accumulated enrich full and accurate such as reconnoitring, the data such as earthquake, well logging.These are containing abundant geological information with different structure, the data that are stored in diverse location, have obvious potential using value, are the important evidence of the following exploration and development of oil gas field, are important sci-tech innovation resources.Utilize method and the technology such as Knowledge Discovery, artificial intelligence, from the existing oil-gas exploration data that pile up like a mountain, find out effective, novel, that there is potential utility, final intelligible rule and pattern, and integrate the large scale scale heterogeneous data resource of this class and realize data semantic and share, can provide more complete and reliable data, services support for decision maker, thereby instruct the technology decision-making of oil-gas exploration enterprise formulation science, for enterprise creates huge economic benefit.This is from " management data ", to rise to the effective way of " managerial knowledge ", is also important topic and the direction of current and following each field in-depth Information System configuration development.

The informatization of petroleum exploration domain starts from six the seventies, because the specialized fields relating to is many, each specialty all adopts the standard of sealing, do not have common standard to follow, therefore caused the phenomenon that between different departments and system, information sharing difficulty, a large amount of " information island " exist, for the integrated application of petroleum exploration domain data with share and brought challenge.For the difficulty facing in oil-gas exploration Data management and utilizaition, Chinese scholars has proposed some active data integrated approaches and has developed corresponding system.But these methods have still adopted the integrated technology of traditional database and middleware Technology generally, lack real-time semantic support, at the aspects such as dynamic mapping of inquiry request semantic extension, resource, have many limitations, and accuracy rate and efficiency are not satisfactory.

The data relevant to oil-gas exploration comprise underground tectonic structure, lithology, physical property, electrically, the raw storage lid of oil gas etc.These data are from different departments and area, various informative (comprising the image, video, text of relational database, GIS data, CAD figure, various forms etc.), and there is different structure type (comprise structurized, semi-structured with non-structured), be difficult to the multi-source heterogeneous data of this class to manage uniformly integrated and shared with semantic hierarchies.The more important thing is, increasing hydrocarbon resources can the exploration information based on existing be explored and develop, they have become the important data assets in oil company, to improving accuracy and the correct decision-making of auxiliary formulation of the oil-gas exploration level of information technology, data interpretation, all significant, effective utilization of these data, also has important effect to improving the economic benefit of oil company.In addition, the change of decision maker's mode of thinking is also for the knowledge architecture of data has been established ideal basis, the management philosophy of " buy but not build " is advocated in modern data management, i.e. competition is not placed on and how obtains and to occupy in data, but is placed in the careful analysis of data and Study on Interpretation.Therefore, from these multi-source heterogeneous oil-gas exploration data, carrying out the unified of knowledge builds and semantic integrated having great importance.

Body, as a kind of clear and definite Formal Specification of shared ideas model, has good concept hierarchy and reasoning from logic tenability, be applicable to describe very much domain knowledge standard, and to the semanteme of isomeric data, integrated and inquiry has very strong supporting role.Body is introduced in oil-gas exploration data integration, contributed to solve the difficulties that it lacks the aspects such as semantic support and standardization in data integration.Therefore, ontology is incorporated into oil-gas exploration Knowledge Discovery and knowledge engineering field, becoming " data management " is " information management ", contributes to solve the problem that the aspects such as current oil-gas exploration data sharing degree is low, Semantic Heterogeneous exist.In order to seek the breakthrough in the integrated theory of oil-gas exploration data semantic and method, oil-gas exploration data semantic integration mechanism and model that foundation builds based on domain body, can effectively improve the value of available data, the science decision of petroleum engineering relevant departments will be had to important theory significance and practical value.

Summary of the invention

The object of the embodiment of the present invention is to provide a kind of integrated models and methods of multi-source heterogeneous data semantic building based on domain body, is intended to solve the integrated and shared problem of semanteme between oil-gas exploration isomeric data.

The embodiment of the present invention is achieved in that a kind of integrated method of multi-source heterogeneous data semantic building based on domain body, and the integrated method of multi-source heterogeneous data semantic that should build based on domain body comprises:

Utilize ontology construction, automatically, semi-automatically set up the local ontology of multi-source heterogeneous data, by ontology merging technique construction domain body, and set up the Semantic mapping relation between data source and local ontology, local ontology and domain body;

In conjunction with society's mark and the complementary advantage of body in knowledge representation, stipulations and expansion are inquired about in semantic query request to user, and the semantic query statement of generating standard, inquires about respectively a plurality of data sources, then by Query Result duplicate removal and optimizing polymerization, finally return to user.

Another object of the present invention is to provide a kind of integrated model of multi-source heterogeneous data semantic building based on domain body, the multi-source heterogeneous data semantic integrated model that should build based on domain body comprises: local ontology builds module, ontology merging module and semantic query dynamic expansion and stipulations module;

Local ontology builds module, according to data source feature, selects adaptively body construction strategy, thereby constructs oil-gas exploration local ontology;

Ontology merging module, building module with local ontology is connected, the ontology merging method that employing combines concept matching and attributes match, utilize maximum information coefficient (MIC) method to calculate the semantic similarity of Concept Semantic Similarity and concept attribute, realize a plurality of local ontology to the flexible merging of domain body;

Semantic query dynamic expansion and stipulations module, build module with local ontology and be connected, for the optimizing polymerization of validity and the result of inquiry request dynamic expansion.

Further, local ontology builds module, according to data source feature, by self-adaptation body construction strategy, carries out the structure of local ontology, specifically comprises:

Step 1, based on unstructured data sources, build local ontology:

First, applicating text filtrator changes into different file layouts into text-only file form, obtains language material data, and carries out consistency check; Then, adopt reverse maximum classification Chinese word cutting method to carry out preliminary cutting to these language materials and process, obtain word string set; Then, utilize maximum information coefficient (MIC) method to calculate the internal bond strength of word string, obtain compound word set, and judge the domain-specific of compound word and non-compound word, extract concept set; Then, the classification relation on application drawing between Random Walk Algorithm fuzzy filtering word concept, adopts the clustering algorithm based on hidden Markov model (HMM) to extract the classification relation between non-compound word concept; Then, use the method based on association rule mining to obtain the non-categorical relation between concept; Finally, the local ontology of applied ontology the build tool output OWL form;

Step 2, builds local ontology based on structured data source:

First, utilize the Semantic mapping relation between R2O technology building database pattern and ontology model, thereby be the concept in body the relationship map in relational database, attribute is mapped as to OWL attribute accordingly, and the relation table of database is converted into body class, the data in database are converted into example; Then, the initial local ontology extracting from database is done to a series of standardization work, by carrying out semantic similarity calculating with standard body, the ontology information that meets threshold value is set up to semantic relation, the ontology information that does not meet threshold value carries out standardization processing, thereby constructs satisfactory standardization local ontology;

Step 3, builds local ontology based on semi-structured data source

Due to semi-structured data between structuring and unstructured data, there is implicit structure but lack class data of fixing or strict structure, so, ontology construction based on above-mentioned two kinds of data types also can be applied to semi-structured data source, first, extract semi-structured data pattern, given mapping ruler, utilizes XML2RD method, and semi-structured data is converted into structural data; Then, according to structural data, build the local ontology corresponding to method construct semi-structured data source of local ontology.

Further, the method for ontology merging block merging is:

The ontology merging method that employing combines concept matching and attributes match, utilize maximum information coefficient MIC method to calculate the semantic similarity of Concept Semantic Similarity and concept attribute, then, by similarity assessment function, the similarity between concept is assessed, output similar matrix, and use field axiom constraint knowledge further to assess its similarity to similar matrix; Then,, by the method training study sorter of machine learning, utilize learning classification device to calculate the similarity between concept example; Finally, by in conjunction with ISO15926 oil gas body and fuzzy formal concept analysis method, consider symmetry and the transitive relations of semantic similarity, fuzzy set theory is introduced in the setting of semantic similarity, realize a plurality of local ontology to the flexible merging of domain body.

Further, the concrete grammar that semantic query dynamic expansion and stipulations module realize is:

First, the conceptual relation and the inferential capability that by the semantic analysis of society mark and body, comprise, inquiry request is carried out to grammer and stipulations semantically and expansion, the semantic query statement of generating standard, solve between inquiry request and domain body data source different the caused mismatch problems due to expression-form, and automatically recommend the semantic respective labels of cluster according to user's inquiry request, for realizing data source, accurately assemble guiding is provided; Then, by the semantic similarity calculating between expanding query request and domain body concept, quantize the degree of association between request and resource concept; Finally, the Concept Semantic relation of enriching of utilizing society's mark and body to comprise, Query Result pattern is carried out to semantic annotations, according to the semanteme overall situation effect of society's mark, introducing is usingd the most relevant credible mark that statistic analysis result obtains data source pointed as one of Query Result reliability evaluation standard, result set is carried out to duplicate removal and optimizing polymerization, realize believable Top-K inquiry.

The integrated models and methods of multi-source heterogeneous data semantic building based on domain body provided by the invention, by structure and mapping, query semantics expansion and the result optimizing polymerization of domain body, provide a kind of oil-gas exploration isomeric data building based on domain body semantic integrated model, solved integrated and shared between oil-gas exploration isomeric data, this semanteme integrated model also has loose coupling, easily expansion, supports the superperformances such as semantic query.

The present invention has realized the dynamic growth of data source, for newly-increased data source, only need to provide corresponding wrapper, builds corresponding local ontology, can improve dirigibility and the practicality of integrated system.With domain body, domain knowledge is described, local ontology is described the Heterogeneous Information knowledge in a certain field, and the mapping of setting up respectively mapping, local ontology and the data source of domain body and local ontology, domain body, local ontology and data source were both interknited, relatively independent again, can reduce the coupling of semantic integrated system.In order to realize semantic query and ease for use, in conjunction with society's mark and the complementary advantage of body in knowledge representation, stipulations and expansion are inquired about in semantic query request to user, and to Query Result duplicate removal and optimizing polymerization, the result after optimizing the most at last returns to user.

Accompanying drawing explanation

Fig. 1 is the integrated method flow diagram of multi-source heterogeneous data semantic building based on domain body that the embodiment of the present invention provides;

Fig. 2 is the structural representation of the multi-source heterogeneous data semantic integrated model building based on domain body that provides of the embodiment of the present invention;

In figure: 1, local ontology builds module; 2, ontology merging module; 3, semantic query dynamic expansion and stipulations module;

Fig. 3 is the embodiment of the multi-source heterogeneous data semantic integrated model building based on domain body that provides of the embodiment of the present invention;

Fig. 4 is the process flow diagram that the unstructured data sources that provides of the embodiment of the present invention builds local ontology;

Fig. 5 is the simple abstract body schematic diagram that the embodiment of the present invention provides.

Embodiment

In order to make object of the present invention, technical scheme and advantage clearer, below in conjunction with embodiment, the present invention is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.

Below in conjunction with drawings and the specific embodiments, application principle of the present invention is further described.

As shown in Figure 1, the integrated method of multi-source heterogeneous data semantic building based on domain body of the embodiment of the present invention comprises:

S101: utilize ontology construction, automatically, semi-automatically set up the local ontology of multi-source heterogeneous data, by ontology merging technique construction domain body, and set up the Semantic mapping relation of data source and local ontology, local ontology and domain body;

S102: in conjunction with society's mark and the complementary advantage of body in knowledge representation, stipulations and expansion are inquired about in semantic query request to user, and the semantic query statement of generating standard, inquires about respectively a plurality of data sources, then by Query Result duplicate removal and optimizing polymerization, finally return to user.

As shown in Figure 2, the multi-source heterogeneous data semantic integrated model building based on domain body of the embodiment of the present invention mainly by: local ontology builds module 1, ontology merging module 2 and semantic query dynamic expansion and stipulations module 3 forms;

Local ontology builds module 1, according to data source feature, selects adaptively body construction strategy, thereby constructs oil-gas exploration local ontology;

Ontology merging module 2, building module 1 with local ontology is connected, the ontology merging method that employing combines concept matching and attributes match, utilize maximum information coefficient MIC method to calculate the semantic similarity of Concept Semantic Similarity and concept attribute, realize a plurality of local ontology to the flexible merging of domain body;

Semantic query dynamic expansion and stipulations module 3, build module 1 with local ontology and be connected, for the optimizing polymerization of validity and the result of inquiry request dynamic expansion.

3 pairs of multi-source heterogeneous data semantic integrated models that build based on domain body of the present invention are described further by reference to the accompanying drawings:

1, local ontology builds module:

At present, also there is no maturation, unified methodology instructs the structure of body, set up relatively complete domain body, not only need to catch a large amount of field concept knowledge, also need to solve the semantic conflict of these field concepts, the problems such as ambiguity, this work that causes body to build is both dull, difficult thorny again, become the bottleneck of knowledge acquisition, in addition, although the editing environment of existing body the build tool can meet the needs of setting up body, but by the manual relation of collecting between field concept and field concept, build body completely, remain a job of wasting time and energy, make the application based on body be difficult to promote, therefore, research is automatic from multi-source heterogeneous data, semi-automatically build domain body, and realize the fast mapping of domain body and multi-source heterogeneous data, the semanteme during thereby solution body builds and inquires about is inconsistent and time bottleneck problem, this has important theory significance and a researching value to the data semantic based on domain body is integrated,

Petroleum exploration domain body self-adaptation construction strategy can be according to data source feature, select adaptively different body construction strategy, be mainly divided into four aspects: based on unstructured data sources (as document, webpage etc.), build local ontology, based on structured data source (as relational database), build local ontology, based on semi-structured data source structure local ontology and local ontology, merge into domain body.

1) based on unstructured data sources, build local ontology:

When unstructured data sources is built into body, first to carry out data pre-service, the pretreated fundamental purpose of data is by filtering unworthy symbol and word, thereby extract significant term, current, some natural language processing instruments (as OpenNLP, CKIP etc.) can be realized this function preferably, in the present invention, adopt OpenNLP to realize the pre-service work of unstructured data sources.

For unstructured data sources, first, applicating text filtrator changes into different file layouts into text-only file form, obtains language material data, and carries out consistency check; Then, adopt reverse maximum classification Chinese word cutting method to carry out preliminary cutting to these language materials and process, obtain word string set; Then, utilize maximum information coefficient (MIC) to calculate the internal bond strength of word string, obtain compound word set, and judge the domain-specific of compound word and non-compound word, extract concept set; Then, the classification relation on application drawing between Random Walk Algorithm fuzzy filtering word concept, adopts the clustering algorithm based on hidden Markov model (HMM) to extract the classification relation between non-compound word concept; Then, use the method based on association rule mining to obtain the non-categorical relation between concept; Finally, the local ontology of applied ontology the build tool output OWL form.Idiographic flow as shown in Figure 4.

2) based on structured data source, build local ontology:

From structured data source, building local ontology mainly considers by Semantic mapping, automatically to build petroleum exploration domain body from relational database.First, utilize the Semantic mapping relation between R2O technology building database pattern and ontology model, thereby be the concept in body the relationship map in relational database, attribute is mapped as to OWL attribute accordingly, and the relation table of database is converted into body class, the data in database are converted into example; Then, the initial local ontology extracting from database is done to standardization work, by carrying out semantic similarity calculating with standard body, the ontology information that meets threshold value is set up to semantic relation, the ontology information that does not meet threshold value carries out standardization processing, thereby constructs normalized local ontology.

3) based on semi-structured data source, build local ontology:

First, extract semi-structured data pattern, given mapping ruler, utilizes XML2RD method, and semi-structured data is converted into structural data; Then, according to structural data, build the local ontology corresponding to method construct semi-structured data source of local ontology.

2, ontology merging module:

The research of ontology merging at present mainly concentrates in concept matching, by coupling, find two semantic corresponding relations between Ontological concept, this will cause the incomplete situation of ontology merging result, in the present invention, adopt the ontology merging method that concept matching and attributes match are combined, utilize maximum information coefficient MIC method to calculate the semantic similarity of Concept Semantic Similarity and concept attribute, then, by similarity assessment function, the similarity between concept is assessed, output similar matrix, and use field axiom constraint knowledge further to assess its similarity to similar matrix, then,, by the method training study sorter of machine learning, utilize learning classification device to calculate the similarity between concept example, finally, by in conjunction with ISO15926 oil gas body and fuzzy formal concept analysis (FFCA) method, consider symmetry and the transitive relations of semantic similarity, fuzzy set theory is introduced in the setting of semantic similarity, thereby realize a plurality of local ontology to the flexible merging of domain body,

3, semantic query dynamic expansion and stipulations module:

The validity of inquiry request dynamic expansion and the optimizing polymerization of result are the integrated another one key issues of oil-gas exploration data semantic, first, the conceptual relation and the inferential capability that by the semantic analysis of society mark and body, comprise, inquiry request is carried out to grammer and stipulations semantically and expansion, the semantic query statement of generating standard (normally SPARQL statement), solve between inquiry request and domain body data source different the caused mismatch problems due to expression-form, and automatically recommend the semantic respective labels of cluster according to user's inquiry request, for realizing data source, accurately focus on guiding is provided, then, by the semantic similarity calculating between expanding query request and domain body concept, quantize the degree of association between request and resource concept, finally, the Concept Semantic relation of enriching of utilizing society's mark and body to comprise, Query Result pattern is carried out to semantic annotations, according to the semanteme overall situation effect of society's mark, introducing is usingd the most relevant credible mark that statistic analysis result obtains data source pointed as one of Query Result reliability evaluation standard, result set is carried out to duplicate removal and optimizing polymerization, realize believable Top-K inquiry.

Semantic distance is often referred to two close degree of the semanteme between concept, it is a kind of effective means of weighing semantic similarity, semantic distance between two concepts is less, their semanteme is more approaching, otherwise far away, semantic distance and semantic similarity are generally inverse relation, in information retrieval field, semantic distance is less, descriptive information is more approaching with user's inquiry request, when semantic distance is 0, information meets user's request completely, when semantic distance is greater than certain value, information and user inquire about onrelevant, therefore can not as a result of collect and return, for the result set returning, whether the arbitrary result in the own subjective judgement result set of user meets the demands completely, semantic distance based on body is mainly measured in body the length of fillet between concept, two concepts access path in body is shorter, they are just more similar, the quantity of information that the length of every fillet is comprised by it determines, if p (c) is the probability of happening of concept c in whole concept set, if count (c) is the occurrence number of concept c in body O, count (O) is the total number of concept in body O, because concept may appear at the form of different abstraction hierarchies in body, the occurrence number of its all sub-concepts that should add up while calculating the total occurrence number of concept, the computing method of p (c) are:

$p\left(c\right)=\frac{\mathrm{count}\left(c\right)+\underset{H(c^{\′},c)}{\mathrm{\Σ}}\mathrm{count}\left(c^{\′}\right)}{\mathrm{count}\left(O\right)}---\left(1\right)$

From formula (1), p (c) is along with the rising monotone increasing of c place level in body O, and p (Root)=1, as shown in Figure 5, and p (c ₀)=1, p (c ₁)=0.5;

If parent (c) is the father set of concept c in body O:

According to the acyclicity of H relation, easily know:

$\∀c\∈C,\mathrm{if\; parent}\left(c\right)\≠\mathrm{\φ},\mathrm{then}\left|\mathrm{parent}\left(c\right)\right|=1---\left(3\right)$

Wherein, | parent (c) | represent to gather the element number comprising in parent (c);

From information theory, if the frequency of a concept appearance is larger, the quantity of information that it comprises is just fewer; Otherwise if the frequency of a concept appearance is less, the quantity of information that it comprises is just more, therefore, the quantity of information that concept c comprises is:

I(c)＝-log(p(c)) (4)

Therefore the quantity of information that, fillet c → parent (c) comprises is:

$I(c\→\mathrm{parent}\left(e\right))=-\log \left(p(c\→\mathrm{parent}\left(c\right))\right)=-\log \left(p\left(c\right|\mathrm{parent}\left(c\right))\right)$

$=-\log \left(\frac{p\left(c\right)}{p\left(\mathrm{parent}\left(c\right)\right)}\right)=I\left(c\right)-I\left(\mathrm{parent}\left(c\right)\right)---\left(5\right)$

The length of fillet c → parent (c) is proportional to the quantity of information that it comprises:

length(c→parent(c))∝I(c→parent(c)) (6)

By formula (5) and formula (6), can be obtained:

length(c→parent(c))＝β·(I(c→parent(c))) (7)

Wherein, β is constant, is a scale factor, in order to express conveniently, makes β=1, can obtain:

length(c→parent(c))＝I(c→parent(c)) (8)

Due to set up given any two concept c ₁and c ₂, establish LCA (c ₁, c ₂) be c ₁and c ₂the public ancestors of minimum in body O (Least Common Ancestor), LCA (c ₁, c ₂) be c ₁and c ₂, for any two concept nodes, necessarily there are the public ancestors of minimum in the concept node of depth capacity on common ancestor path, as shown in Figure 5, and LCA (c ₃, c ₉)=c ₁, LCA (c ₁, c ₅)=c ₀, i.e. the root concept node (Root) of body, therefore:

Definition 1 (access path): due to set up, and H relation has acyclicity, c ₁and c ₂access path in body O has and only has one, is designated as path (c ₁, c ₂), and c ₁to LCA (c ₁, c ₂) path and c ₂to LCA (c ₁, c ₂) union in path is access path, be defined as:

path(c ₁,c ₂)＝path(c ₁,LCA(c ₁,c ₂))∪path(c ₂,LCA(c ₁,c ₂)) (10)

Definition 2 (semantic distances): semantic distance has reflected the semantic association degree between two concepts, and the semantic distance between two concepts is defined as:

$\begin{array}{c}\mathrm{dis}(c_1,c_2)=\underset{c\∈\{\mathrm{path}(c_1,c_2)-\mathrm{LCA}(c_1,c_2)\}}{\mathrm{\Σ}}\mathrm{length}(c\→\mathrm{parent}\left(c\right))\\ =\underset{c\∈\{\mathrm{path}(c_1,c_2)-\mathrm{LCA}(c_1,c_2)\}}{\mathrm{\Σ}}I\left(c\right)-I\left(\mathrm{parent}\left(\text{c}\right)\right)\\ =(I\left(c_1\right)-I\left(\mathrm{LCA}(c_1,c_2)\right)+(I\left(c_2\right)-I\left(\mathrm{LCA}(c_1,c_2)\right)\\ =2\log \left(p\left(\mathrm{LCA}(c_1,c_2)\right)\right)-\left(\log \left(p(c_1\right)\right)+\log \left(p\left(c_2\right))\right)\end{array}---\left(11\right)$

From formula (11), in fact the definition of semantic distance is comprised the information of two aspects here: on the one hand, the formation of domain body has determined LCA (c ₁, c ₂) position, this is when measuring similarity, within the inheritance between concept and concept, in body level, residing position has all been included in semantic distance to the impact of similarity; On the other hand, p (LCA (c ₁, c ₂)), p (c ₁) and p (c ₂) value all come from the statistical information of concept set,

From above-mentioned definition, with respect to different domain bodies, semantic distance between concept is different, and different concepts is with respect to same domain body, its semantic distance is also different, this computing method to semantic distance combine semantic relation between concept and the statistical information of objective fact, contribute to simulate more accurately the original appearance of objective world.

The integrated domain body that first need to be participated in creating by domain expert of semanteme of isomeric data of the present invention provides shared knowledge base, next needs ability to express to enrich and have the ontology describing language of certain logical reasoning ability, then by selecting rational mapping method, the Uniform semantic information that becomes integrated system to understand the Semantic Heterogeneous data-switching in different pieces of information source.In addition, system also should be supported semantic query, and has certain ease for use and extensibility.

This model can be realized the dynamic growth of data source, for newly-increased data source, only need to provide corresponding wrapper, builds corresponding local ontology, can improve dirigibility and the practicality of integrated system.With domain body, domain knowledge is described, local ontology is described the Heterogeneous Information knowledge in a certain field, and the mapping of setting up respectively mapping, local ontology and the data source of domain body and local ontology, this both interknited domain body, local ontology and data source, relatively independent again, can reduce the coupling of semantic integrated system.In order to realize semantic query and ease for use, in conjunction with society's mark and the complementary advantage of body in knowledge representation, user's semantic query request is inquired about to stipulations and expansion, and to Query Result duplicate removal and optimizing polymerization, finally return to user.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any modifications of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection scope of the present invention.

Claims (5)

1. the integrated method of multi-source heterogeneous data semantic building based on domain body, is characterized in that, the integrated method of multi-source heterogeneous data semantic that should build based on domain body comprises:

Utilize ontology construction, automatically, semi-automatically set up the local ontology of multi-source heterogeneous data, by ontology merging technique construction domain body, and set up the Semantic mapping relation between data source and local ontology, local ontology and domain body;

In conjunction with society's mark and the complementary advantage of body in knowledge representation, stipulations and expansion are inquired about in semantic query request to user, and the semantic query statement of generating standard, inquires about respectively a plurality of data sources, then by Query Result duplicate removal and optimizing polymerization, finally return to user.

2. the integrated model of multi-source heterogeneous data semantic building based on domain body, it is characterized in that, the multi-source heterogeneous data semantic integrated model that should build based on domain body comprises: local ontology builds module, ontology merging module and semantic query dynamic expansion and stipulations module;

Local ontology builds module, according to data source feature, selects adaptively body construction strategy, thereby constructs oil-gas exploration local ontology;

Ontology merging module, building module with local ontology is connected, the ontology merging method that employing combines concept matching and attributes match, utilize maximum information coefficient method to calculate the semantic similarity of Concept Semantic Similarity and concept attribute, realize a plurality of local ontology to the flexible merging of domain body;

Semantic query dynamic expansion and stipulations module, build module with local ontology and be connected, for the optimizing polymerization of validity and the result of inquiry request dynamic expansion.

3. the integrated model of multi-source heterogeneous data semantic building based on domain body as claimed in claim 2, is characterized in that, local ontology builds module, according to data source feature, by self-adaptation body construction strategy, carries out the structure of local ontology, specifically comprises:

Step 1, based on unstructured data sources, build local ontology:

First, applicating text filtrator changes into different file layouts into text-only file form, obtains language material data, and carries out consistency check; Then, adopt reverse maximum classification Chinese word cutting method to carry out preliminary cutting to these language materials and process, obtain word string set; Then, utilize maximum information coefficient method to calculate the internal bond strength of word string, obtain compound word set, and judge the domain-specific of compound word and non-compound word, extract concept set; Then, the classification relation on application drawing between Random Walk Algorithm fuzzy filtering word concept, adopts the clustering algorithm based on hidden Markov model to extract the classification relation between non-compound word concept; Then, use the method based on association rule mining to obtain the non-categorical relation between concept; Finally, the local ontology of applied ontology the build tool output OWL form;

Step 2, builds local ontology based on structured data source:

First, utilize the Semantic mapping relation between R2O technology building database pattern and ontology model, thereby be the concept in body the relationship map in relational database, attribute is mapped as to OWL attribute accordingly, and the relation table of database is converted into body class, the data in database are converted into example; Then, the initial local ontology extracting from database is done to a series of standardization work, by carrying out semantic similarity calculating with standard body, the ontology information that meets threshold value is set up to semantic relation, the ontology information that does not meet threshold value carries out standardization processing, thereby constructs satisfactory standardization local ontology;

Step 3, builds local ontology based on semi-structured data source

Due to semi-structured data between structuring and unstructured data, there is implicit structure but lack class data of fixing or strict structure; So the ontology construction based on above-mentioned two kinds of data types also can be applied to semi-structured data source; First, extract semi-structured data pattern, given mapping ruler, utilizes XML2RD method, and semi-structured data is converted into structural data; Then, according to structural data, build the local ontology corresponding to method construct semi-structured data source of local ontology.

4. the integrated model of multi-source heterogeneous data semantic building based on domain body as claimed in claim 2, is characterized in that, the method for ontology merging block merging is:

The ontology merging method that employing combines concept matching and attributes match, utilize maximum information coefficient method to calculate the semantic similarity of Concept Semantic Similarity and concept attribute, then, by similarity assessment function, the similarity between concept is assessed, output similar matrix, and use field axiom constraint knowledge further to assess its similarity to similar matrix; Then,, by the method training study sorter of machine learning, utilize learning classification device to calculate the similarity between concept example; Finally, by in conjunction with ISO15926 oil gas body and fuzzy formal concept analysis method, consider symmetry and the transitive relations of semantic similarity, fuzzy set theory is introduced in the setting of semantic similarity, realize a plurality of local ontology to the flexible merging of domain body.

5. the integrated model of multi-source heterogeneous data semantic building based on domain body as claimed in claim 2, is characterized in that, the concrete grammar that semantic query dynamic expansion and stipulations module realize is:

First, the conceptual relation and the inferential capability that by the semantic analysis of society mark and body, comprise, inquiry request is carried out to grammer and stipulations semantically and expansion, the semantic query statement of generating standard, solve between inquiry request and domain body data source different the caused mismatch problems due to expression-form, and automatically recommend the semantic respective labels of cluster according to user's inquiry request, for realizing data source, accurately assemble guiding is provided; Then, by the semantic similarity calculating between expanding query request and domain body concept, quantize the degree of association between request and resource concept; Finally, the Concept Semantic relation of enriching of utilizing society's mark and body to comprise, Query Result pattern is carried out to semantic annotations, according to the semanteme overall situation effect of society's mark, introducing is usingd the most relevant credible mark that statistic analysis result obtains data source pointed as one of Query Result reliability evaluation standard, result set is carried out to duplicate removal and optimizing polymerization, realize believable Top-K inquiry.